branching: merge stable into default

This show that the recent changes on default fixed the issue with transaction
overwriting content in `test-transaction-wc-rollback-race.t`

contrib/setup-pytype.sh

@@ -0,0 +1,33 b''
+#!/bin/bash
+
+set -e
+set -u
+
+# Find the python3 setup that would run pytype
+PYTYPE=`which pytype`
+PYTHON3=`head -n1 ${PYTYPE} | sed -s 's/#!//'`
+
+# Existing stubs that pytype processes live here
+TYPESHED=$(${PYTHON3} -c "import pytype; print(pytype.__path__[0])")/typeshed/stubs
+HG_STUBS=${TYPESHED}/mercurial
+
+echo "Patching typeshed at $HG_STUBS"
+
+rm -rf ${HG_STUBS}
+mkdir -p ${HG_STUBS}
+
+cat > ${HG_STUBS}/METADATA.toml <<EOF
+version = "0.1"
+EOF
+
+
+mkdir -p ${HG_STUBS}/mercurial/cext ${HG_STUBS}/mercurial/thirdparty/attr
+
+touch ${HG_STUBS}/mercurial/__init__.pyi
+touch ${HG_STUBS}/mercurial/cext/__init__.pyi
+touch ${HG_STUBS}/mercurial/thirdparty/__init__.pyi
+
+ln -sf $(hg root)/mercurial/cext/*.{pyi,typed} \
+       ${HG_STUBS}/mercurial/cext
+ln -sf $(hg root)/mercurial/thirdparty/attr/*.{pyi,typed} \
+       ${HG_STUBS}/mercurial/thirdparty/attr

mercurial/cext/py.typed

@@ -0,0 +1,1 b''
+partial

mercurial/thirdparty/attr/__init__.pyi

@@ -0,0 +1,486 b''
+import sys
+
+from typing import (
+    Any,
+    Callable,
+    ClassVar,
+    Dict,
+    Generic,
+    List,
+    Mapping,
+    Optional,
+    Protocol,
+    Sequence,
+    Tuple,
+    Type,
+    TypeVar,
+    Union,
+    overload,
+)
+
+# `import X as X` is required to make these public
+from . import converters as converters
+from . import exceptions as exceptions
+from . import filters as filters
+from . import setters as setters
+from . import validators as validators
+from ._cmp import cmp_using as cmp_using
+from ._version_info import VersionInfo
+
+__version__: str
+__version_info__: VersionInfo
+__title__: str
+__description__: str
+__url__: str
+__uri__: str
+__author__: str
+__email__: str
+__license__: str
+__copyright__: str
+
+_T = TypeVar("_T")
+_C = TypeVar("_C", bound=type)
+
+_EqOrderType = Union[bool, Callable[[Any], Any]]
+_ValidatorType = Callable[[Any, Attribute[_T], _T], Any]
+_ConverterType = Callable[[Any], Any]
+_FilterType = Callable[[Attribute[_T], _T], bool]
+_ReprType = Callable[[Any], str]
+_ReprArgType = Union[bool, _ReprType]
+_OnSetAttrType = Callable[[Any, Attribute[Any], Any], Any]
+_OnSetAttrArgType = Union[
+    _OnSetAttrType, List[_OnSetAttrType], setters._NoOpType
+]
+_FieldTransformer = Callable[
+    [type, List[Attribute[Any]]], List[Attribute[Any]]
+]
+# FIXME: in reality, if multiple validators are passed they must be in a list
+# or tuple, but those are invariant and so would prevent subtypes of
+# _ValidatorType from working when passed in a list or tuple.
+_ValidatorArgType = Union[_ValidatorType[_T], Sequence[_ValidatorType[_T]]]
+
+# A protocol to be able to statically accept an attrs class.
+class AttrsInstance(Protocol):
+    __attrs_attrs__: ClassVar[Any]
+
+# _make --
+
+NOTHING: object
+
+# NOTE: Factory lies about its return type to make this possible:
+# `x: List[int] # = Factory(list)`
+# Work around mypy issue #4554 in the common case by using an overload.
+if sys.version_info >= (3, 8):
+    from typing import Literal
+    @overload
+    def Factory(factory: Callable[[], _T]) -> _T: ...
+    @overload
+    def Factory(
+        factory: Callable[[Any], _T],
+        takes_self: Literal[True],
+    ) -> _T: ...
+    @overload
+    def Factory(
+        factory: Callable[[], _T],
+        takes_self: Literal[False],
+    ) -> _T: ...
+
+else:
+    @overload
+    def Factory(factory: Callable[[], _T]) -> _T: ...
+    @overload
+    def Factory(
+        factory: Union[Callable[[Any], _T], Callable[[], _T]],
+        takes_self: bool = ...,
+    ) -> _T: ...
+
+# Static type inference support via __dataclass_transform__ implemented as per:
+# https://github.com/microsoft/pyright/blob/1.1.135/specs/dataclass_transforms.md
+# This annotation must be applied to all overloads of "define" and "attrs"
+#
+# NOTE: This is a typing construct and does not exist at runtime.  Extensions
+# wrapping attrs decorators should declare a separate __dataclass_transform__
+# signature in the extension module using the specification linked above to
+# provide pyright support.
+def __dataclass_transform__(
+    *,
+    eq_default: bool = True,
+    order_default: bool = False,
+    kw_only_default: bool = False,
+    field_descriptors: Tuple[Union[type, Callable[..., Any]], ...] = (()),
+) -> Callable[[_T], _T]: ...
+
+class Attribute(Generic[_T]):
+    name: str
+    default: Optional[_T]
+    validator: Optional[_ValidatorType[_T]]
+    repr: _ReprArgType
+    cmp: _EqOrderType
+    eq: _EqOrderType
+    order: _EqOrderType
+    hash: Optional[bool]
+    init: bool
+    converter: Optional[_ConverterType]
+    metadata: Dict[Any, Any]
+    type: Optional[Type[_T]]
+    kw_only: bool
+    on_setattr: _OnSetAttrType
+    def evolve(self, **changes: Any) -> "Attribute[Any]": ...
+
+# NOTE: We had several choices for the annotation to use for type arg:
+# 1) Type[_T]
+#   - Pros: Handles simple cases correctly
+#   - Cons: Might produce less informative errors in the case of conflicting
+#     TypeVars e.g. `attr.ib(default='bad', type=int)`
+# 2) Callable[..., _T]
+#   - Pros: Better error messages than #1 for conflicting TypeVars
+#   - Cons: Terrible error messages for validator checks.
+#   e.g. attr.ib(type=int, validator=validate_str)
+#        -> error: Cannot infer function type argument
+# 3) type (and do all of the work in the mypy plugin)
+#   - Pros: Simple here, and we could customize the plugin with our own errors.
+#   - Cons: Would need to write mypy plugin code to handle all the cases.
+# We chose option #1.
+
+# `attr` lies about its return type to make the following possible:
+#     attr()    -> Any
+#     attr(8)   -> int
+#     attr(validator=<some callable>)  -> Whatever the callable expects.
+# This makes this type of assignments possible:
+#     x: int = attr(8)
+#
+# This form catches explicit None or no default but with no other arguments
+# returns Any.
+@overload
+def attrib(
+    default: None = ...,
+    validator: None = ...,
+    repr: _ReprArgType = ...,
+    cmp: Optional[_EqOrderType] = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    metadata: Optional[Mapping[Any, Any]] = ...,
+    type: None = ...,
+    converter: None = ...,
+    factory: None = ...,
+    kw_only: bool = ...,
+    eq: Optional[_EqOrderType] = ...,
+    order: Optional[_EqOrderType] = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+) -> Any: ...
+
+# This form catches an explicit None or no default and infers the type from the
+# other arguments.
+@overload
+def attrib(
+    default: None = ...,
+    validator: Optional[_ValidatorArgType[_T]] = ...,
+    repr: _ReprArgType = ...,
+    cmp: Optional[_EqOrderType] = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    metadata: Optional[Mapping[Any, Any]] = ...,
+    type: Optional[Type[_T]] = ...,
+    converter: Optional[_ConverterType] = ...,
+    factory: Optional[Callable[[], _T]] = ...,
+    kw_only: bool = ...,
+    eq: Optional[_EqOrderType] = ...,
+    order: Optional[_EqOrderType] = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+) -> _T: ...
+
+# This form catches an explicit default argument.
+@overload
+def attrib(
+    default: _T,
+    validator: Optional[_ValidatorArgType[_T]] = ...,
+    repr: _ReprArgType = ...,
+    cmp: Optional[_EqOrderType] = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    metadata: Optional[Mapping[Any, Any]] = ...,
+    type: Optional[Type[_T]] = ...,
+    converter: Optional[_ConverterType] = ...,
+    factory: Optional[Callable[[], _T]] = ...,
+    kw_only: bool = ...,
+    eq: Optional[_EqOrderType] = ...,
+    order: Optional[_EqOrderType] = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+) -> _T: ...
+
+# This form covers type=non-Type: e.g. forward references (str), Any
+@overload
+def attrib(
+    default: Optional[_T] = ...,
+    validator: Optional[_ValidatorArgType[_T]] = ...,
+    repr: _ReprArgType = ...,
+    cmp: Optional[_EqOrderType] = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    metadata: Optional[Mapping[Any, Any]] = ...,
+    type: object = ...,
+    converter: Optional[_ConverterType] = ...,
+    factory: Optional[Callable[[], _T]] = ...,
+    kw_only: bool = ...,
+    eq: Optional[_EqOrderType] = ...,
+    order: Optional[_EqOrderType] = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+) -> Any: ...
+@overload
+def field(
+    *,
+    default: None = ...,
+    validator: None = ...,
+    repr: _ReprArgType = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    metadata: Optional[Mapping[Any, Any]] = ...,
+    converter: None = ...,
+    factory: None = ...,
+    kw_only: bool = ...,
+    eq: Optional[bool] = ...,
+    order: Optional[bool] = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+) -> Any: ...
+
+# This form catches an explicit None or no default and infers the type from the
+# other arguments.
+@overload
+def field(
+    *,
+    default: None = ...,
+    validator: Optional[_ValidatorArgType[_T]] = ...,
+    repr: _ReprArgType = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    metadata: Optional[Mapping[Any, Any]] = ...,
+    converter: Optional[_ConverterType] = ...,
+    factory: Optional[Callable[[], _T]] = ...,
+    kw_only: bool = ...,
+    eq: Optional[_EqOrderType] = ...,
+    order: Optional[_EqOrderType] = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+) -> _T: ...
+
+# This form catches an explicit default argument.
+@overload
+def field(
+    *,
+    default: _T,
+    validator: Optional[_ValidatorArgType[_T]] = ...,
+    repr: _ReprArgType = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    metadata: Optional[Mapping[Any, Any]] = ...,
+    converter: Optional[_ConverterType] = ...,
+    factory: Optional[Callable[[], _T]] = ...,
+    kw_only: bool = ...,
+    eq: Optional[_EqOrderType] = ...,
+    order: Optional[_EqOrderType] = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+) -> _T: ...
+
+# This form covers type=non-Type: e.g. forward references (str), Any
+@overload
+def field(
+    *,
+    default: Optional[_T] = ...,
+    validator: Optional[_ValidatorArgType[_T]] = ...,
+    repr: _ReprArgType = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    metadata: Optional[Mapping[Any, Any]] = ...,
+    converter: Optional[_ConverterType] = ...,
+    factory: Optional[Callable[[], _T]] = ...,
+    kw_only: bool = ...,
+    eq: Optional[_EqOrderType] = ...,
+    order: Optional[_EqOrderType] = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+) -> Any: ...
+@overload
+@__dataclass_transform__(order_default=True, field_descriptors=(attrib, field))
+def attrs(
+    maybe_cls: _C,
+    these: Optional[Dict[str, Any]] = ...,
+    repr_ns: Optional[str] = ...,
+    repr: bool = ...,
+    cmp: Optional[_EqOrderType] = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    slots: bool = ...,
+    frozen: bool = ...,
+    weakref_slot: bool = ...,
+    str: bool = ...,
+    auto_attribs: bool = ...,
+    kw_only: bool = ...,
+    cache_hash: bool = ...,
+    auto_exc: bool = ...,
+    eq: Optional[_EqOrderType] = ...,
+    order: Optional[_EqOrderType] = ...,
+    auto_detect: bool = ...,
+    collect_by_mro: bool = ...,
+    getstate_setstate: Optional[bool] = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+    field_transformer: Optional[_FieldTransformer] = ...,
+    match_args: bool = ...,
+) -> _C: ...
+@overload
+@__dataclass_transform__(order_default=True, field_descriptors=(attrib, field))
+def attrs(
+    maybe_cls: None = ...,
+    these: Optional[Dict[str, Any]] = ...,
+    repr_ns: Optional[str] = ...,
+    repr: bool = ...,
+    cmp: Optional[_EqOrderType] = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    slots: bool = ...,
+    frozen: bool = ...,
+    weakref_slot: bool = ...,
+    str: bool = ...,
+    auto_attribs: bool = ...,
+    kw_only: bool = ...,
+    cache_hash: bool = ...,
+    auto_exc: bool = ...,
+    eq: Optional[_EqOrderType] = ...,
+    order: Optional[_EqOrderType] = ...,
+    auto_detect: bool = ...,
+    collect_by_mro: bool = ...,
+    getstate_setstate: Optional[bool] = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+    field_transformer: Optional[_FieldTransformer] = ...,
+    match_args: bool = ...,
+) -> Callable[[_C], _C]: ...
+@overload
+@__dataclass_transform__(field_descriptors=(attrib, field))
+def define(
+    maybe_cls: _C,
+    *,
+    these: Optional[Dict[str, Any]] = ...,
+    repr: bool = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    slots: bool = ...,
+    frozen: bool = ...,
+    weakref_slot: bool = ...,
+    str: bool = ...,
+    auto_attribs: bool = ...,
+    kw_only: bool = ...,
+    cache_hash: bool = ...,
+    auto_exc: bool = ...,
+    eq: Optional[bool] = ...,
+    order: Optional[bool] = ...,
+    auto_detect: bool = ...,
+    getstate_setstate: Optional[bool] = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+    field_transformer: Optional[_FieldTransformer] = ...,
+    match_args: bool = ...,
+) -> _C: ...
+@overload
+@__dataclass_transform__(field_descriptors=(attrib, field))
+def define(
+    maybe_cls: None = ...,
+    *,
+    these: Optional[Dict[str, Any]] = ...,
+    repr: bool = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    slots: bool = ...,
+    frozen: bool = ...,
+    weakref_slot: bool = ...,
+    str: bool = ...,
+    auto_attribs: bool = ...,
+    kw_only: bool = ...,
+    cache_hash: bool = ...,
+    auto_exc: bool = ...,
+    eq: Optional[bool] = ...,
+    order: Optional[bool] = ...,
+    auto_detect: bool = ...,
+    getstate_setstate: Optional[bool] = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+    field_transformer: Optional[_FieldTransformer] = ...,
+    match_args: bool = ...,
+) -> Callable[[_C], _C]: ...
+
+mutable = define
+frozen = define  # they differ only in their defaults
+
+def fields(cls: Type[AttrsInstance]) -> Any: ...
+def fields_dict(cls: Type[AttrsInstance]) -> Dict[str, Attribute[Any]]: ...
+def validate(inst: AttrsInstance) -> None: ...
+def resolve_types(
+    cls: _C,
+    globalns: Optional[Dict[str, Any]] = ...,
+    localns: Optional[Dict[str, Any]] = ...,
+    attribs: Optional[List[Attribute[Any]]] = ...,
+) -> _C: ...
+
+# TODO: add support for returning a proper attrs class from the mypy plugin
+# we use Any instead of _CountingAttr so that e.g. `make_class('Foo',
+# [attr.ib()])` is valid
+def make_class(
+    name: str,
+    attrs: Union[List[str], Tuple[str, ...], Dict[str, Any]],
+    bases: Tuple[type, ...] = ...,
+    repr_ns: Optional[str] = ...,
+    repr: bool = ...,
+    cmp: Optional[_EqOrderType] = ...,
+    hash: Optional[bool] = ...,
+    init: bool = ...,
+    slots: bool = ...,
+    frozen: bool = ...,
+    weakref_slot: bool = ...,
+    str: bool = ...,
+    auto_attribs: bool = ...,
+    kw_only: bool = ...,
+    cache_hash: bool = ...,
+    auto_exc: bool = ...,
+    eq: Optional[_EqOrderType] = ...,
+    order: Optional[_EqOrderType] = ...,
+    collect_by_mro: bool = ...,
+    on_setattr: Optional[_OnSetAttrArgType] = ...,
+    field_transformer: Optional[_FieldTransformer] = ...,
+) -> type: ...
+
+# _funcs --
+
+# TODO: add support for returning TypedDict from the mypy plugin
+# FIXME: asdict/astuple do not honor their factory args. Waiting on one of
+# these:
+# https://github.com/python/mypy/issues/4236
+# https://github.com/python/typing/issues/253
+# XXX: remember to fix attrs.asdict/astuple too!
+def asdict(
+    inst: AttrsInstance,
+    recurse: bool = ...,
+    filter: Optional[_FilterType[Any]] = ...,
+    dict_factory: Type[Mapping[Any, Any]] = ...,
+    retain_collection_types: bool = ...,
+    value_serializer: Optional[
+        Callable[[type, Attribute[Any], Any], Any]
+    ] = ...,
+    tuple_keys: Optional[bool] = ...,
+) -> Dict[str, Any]: ...
+
+# TODO: add support for returning NamedTuple from the mypy plugin
+def astuple(
+    inst: AttrsInstance,
+    recurse: bool = ...,
+    filter: Optional[_FilterType[Any]] = ...,
+    tuple_factory: Type[Sequence[Any]] = ...,
+    retain_collection_types: bool = ...,
+) -> Tuple[Any, ...]: ...
+def has(cls: type) -> bool: ...
+def assoc(inst: _T, **changes: Any) -> _T: ...
+def evolve(inst: _T, **changes: Any) -> _T: ...
+
+# _config --
+
+def set_run_validators(run: bool) -> None: ...
+def get_run_validators() -> bool: ...
+
+# aliases --
+
+s = attributes = attrs
+ib = attr = attrib
+dataclass = attrs  # Technically, partial(attrs, auto_attribs=True) ;)

mercurial/thirdparty/attr/_cmp.py

@@ -0,0 +1,155 b''
+# SPDX-License-Identifier: MIT
+
+
+import functools
+import types
+
+from ._make import _make_ne
+
+
+_operation_names = {"eq": "==", "lt": "<", "le": "<=", "gt": ">", "ge": ">="}
+
+
+def cmp_using(
+    eq=None,
+    lt=None,
+    le=None,
+    gt=None,
+    ge=None,
+    require_same_type=True,
+    class_name="Comparable",
+):
+    """
+    Create a class that can be passed into `attr.ib`'s ``eq``, ``order``, and
+    ``cmp`` arguments to customize field comparison.
+
+    The resulting class will have a full set of ordering methods if
+    at least one of ``{lt, le, gt, ge}`` and ``eq``  are provided.
+
+    :param Optional[callable] eq: `callable` used to evaluate equality
+        of two objects.
+    :param Optional[callable] lt: `callable` used to evaluate whether
+        one object is less than another object.
+    :param Optional[callable] le: `callable` used to evaluate whether
+        one object is less than or equal to another object.
+    :param Optional[callable] gt: `callable` used to evaluate whether
+        one object is greater than another object.
+    :param Optional[callable] ge: `callable` used to evaluate whether
+        one object is greater than or equal to another object.
+
+    :param bool require_same_type: When `True`, equality and ordering methods
+        will return `NotImplemented` if objects are not of the same type.
+
+    :param Optional[str] class_name: Name of class. Defaults to 'Comparable'.
+
+    See `comparison` for more details.
+
+    .. versionadded:: 21.1.0
+    """
+
+    body = {
+        "__slots__": ["value"],
+        "__init__": _make_init(),
+        "_requirements": [],
+        "_is_comparable_to": _is_comparable_to,
+    }
+
+    # Add operations.
+    num_order_functions = 0
+    has_eq_function = False
+
+    if eq is not None:
+        has_eq_function = True
+        body["__eq__"] = _make_operator("eq", eq)
+        body["__ne__"] = _make_ne()
+
+    if lt is not None:
+        num_order_functions += 1
+        body["__lt__"] = _make_operator("lt", lt)
+
+    if le is not None:
+        num_order_functions += 1
+        body["__le__"] = _make_operator("le", le)
+
+    if gt is not None:
+        num_order_functions += 1
+        body["__gt__"] = _make_operator("gt", gt)
+
+    if ge is not None:
+        num_order_functions += 1
+        body["__ge__"] = _make_operator("ge", ge)
+
+    type_ = types.new_class(
+        class_name, (object,), {}, lambda ns: ns.update(body)
+    )
+
+    # Add same type requirement.
+    if require_same_type:
+        type_._requirements.append(_check_same_type)
+
+    # Add total ordering if at least one operation was defined.
+    if 0 < num_order_functions < 4:
+        if not has_eq_function:
+            # functools.total_ordering requires __eq__ to be defined,
+            # so raise early error here to keep a nice stack.
+            raise ValueError(
+                "eq must be define is order to complete ordering from "
+                "lt, le, gt, ge."
+            )
+        type_ = functools.total_ordering(type_)
+
+    return type_
+
+
+def _make_init():
+    """
+    Create __init__ method.
+    """
+
+    def __init__(self, value):
+        """
+        Initialize object with *value*.
+        """
+        self.value = value
+
+    return __init__
+
+
+def _make_operator(name, func):
+    """
+    Create operator method.
+    """
+
+    def method(self, other):
+        if not self._is_comparable_to(other):
+            return NotImplemented
+
+        result = func(self.value, other.value)
+        if result is NotImplemented:
+            return NotImplemented
+
+        return result
+
+    method.__name__ = "__%s__" % (name,)
+    method.__doc__ = "Return a %s b.  Computed by attrs." % (
+        _operation_names[name],
+    )
+
+    return method
+
+
+def _is_comparable_to(self, other):
+    """
+    Check whether `other` is comparable to `self`.
+    """
+    for func in self._requirements:
+        if not func(self, other):
+            return False
+    return True
+
+
+def _check_same_type(self, other):
+    """
+    Return True if *self* and *other* are of the same type, False otherwise.
+    """
+    return other.value.__class__ is self.value.__class__

mercurial/thirdparty/attr/_cmp.pyi

@@ -0,0 +1,13 b''
+from typing import Any, Callable, Optional, Type
+
+_CompareWithType = Callable[[Any, Any], bool]
+
+def cmp_using(
+    eq: Optional[_CompareWithType],
+    lt: Optional[_CompareWithType],
+    le: Optional[_CompareWithType],
+    gt: Optional[_CompareWithType],
+    ge: Optional[_CompareWithType],
+    require_same_type: bool,
+    class_name: str,
+) -> Type: ...

.gitlab/merge_request_templates/Default.md

@@ -1,5 +1,8 b''
 /assign_reviewer @mercurial.review
 
+
+<!--
+
 Welcome to the Mercurial Merge Request creation process:
 
 * Set a simple title for your MR,
@@ -11,3 +14,5 b' More details here:'
 
 * https://www.mercurial-scm.org/wiki/ContributingChanges
 * https://www.mercurial-scm.org/wiki/Heptapod
+
+-->

Makefile

@@ -138,6 +138,7 b' tests:'
         # Run Rust tests if cargo is installed
 	if command -v $(CARGO) >/dev/null 2>&1; then \
 		$(MAKE) rust-tests; \
+		$(MAKE) cargo-clippy; \
 	fi
 	cd tests && $(PYTHON) run-tests.py $(TESTFLAGS)
 
@@ -152,9 +153,13 b' testpy-%:'
 	cd tests && $(HGPYTHONS)/$*/bin/python run-tests.py $(TESTFLAGS)
 
 rust-tests:
-	cd $(HGROOT)/rust/hg-cpython \
+	cd $(HGROOT)/rust \
 		&& $(CARGO) test --quiet --all --features "$(HG_RUST_FEATURES)"
 
+cargo-clippy:
+	cd $(HGROOT)/rust \
+		&& $(CARGO) clippy --all --features "$(HG_RUST_FEATURES)" -- -D warnings
+
 check-code:
 	hg manifest | xargs python contrib/check-code.py
 

contrib/check-code.py

@@ -372,10 +372,6 b' commonpypats = ['
         ),
         (r'[^^+=*/!<>&| %-](\s=|=\s)[^= ]', "wrong whitespace around ="),
         (
-            r'\([^()]*( =[^=]|[^<>!=]= )',
-            "no whitespace around = for named parameters",
-        ),
-        (
             r'raise [^,(]+, (\([^\)]+\)|[^,\(\)]+)$',
             "don't use old-style two-argument raise, use Exception(message)",
         ),

contrib/check-pytype.sh

@@ -12,6 +12,36 b' cd `hg root`'
 # endeavor to empty this list out over time, as some of these are
 # probably hiding real problems.
 #
+# hgext/absorb.py               # [attribute-error]
+# hgext/bugzilla.py             # [pyi-error], [attribute-error]
+# hgext/convert/bzr.py          # [attribute-error]
+# hgext/convert/cvs.py          # [attribute-error], [wrong-arg-types]
+# hgext/convert/cvsps.py        # [attribute-error]
+# hgext/convert/p4.py           # [wrong-arg-types] (__file: mercurial.utils.procutil._pfile -> IO)
+# hgext/convert/subversion.py   # [attribute-error], [name-error], [pyi-error]
+# hgext/fastannotate/context.py # no linelog.copyfrom()
+# hgext/fastannotate/formatter.py  # [unsupported-operands]
+# hgext/fsmonitor/__init__.py   # [name-error]
+# hgext/git/__init__.py         # [attribute-error]
+# hgext/githelp.py              # [attribute-error] [wrong-arg-types]
+# hgext/hgk.py                  # [attribute-error]
+# hgext/histedit.py             # [attribute-error], [wrong-arg-types]
+# hgext/infinitepush            # using bytes for str literal; scheduled for removal
+# hgext/keyword.py              # [attribute-error]
+# hgext/largefiles/storefactory.py  # [attribute-error]
+# hgext/lfs/__init__.py         # [attribute-error]
+# hgext/narrow/narrowbundle2.py # [attribute-error]
+# hgext/narrow/narrowcommands.py    # [attribute-error], [name-error]
+# hgext/rebase.py               # [attribute-error]
+# hgext/remotefilelog/basepack.py   # [attribute-error], [wrong-arg-count]
+# hgext/remotefilelog/basestore.py  # [attribute-error]
+# hgext/remotefilelog/contentstore.py   # [missing-parameter], [wrong-keyword-args], [attribute-error]
+# hgext/remotefilelog/fileserverclient.py  # [attribute-error]
+# hgext/remotefilelog/shallowbundle.py     # [attribute-error]
+# hgext/remotefilelog/remotefilectx.py  # [module-attr] (This is an actual bug)
+# hgext/sqlitestore.py          # [attribute-error]
+# hgext/zeroconf/__init__.py    # bytes vs str; tests fail on macOS
+#
 # mercurial/bundlerepo.py       # no vfs and ui attrs on bundlerepo
 # mercurial/context.py          # many [attribute-error]
 # mercurial/crecord.py          # tons of [attribute-error], [module-attr]
@@ -31,7 +61,6 b' cd `hg root`'
 # mercurial/pure/parsers.py     # [attribute-error]
 # mercurial/repoview.py         # [attribute-error]
 # mercurial/testing/storage.py  # tons of [attribute-error]
-# mercurial/ui.py               # [attribute-error], [wrong-arg-types]
 # mercurial/unionrepo.py        # ui, svfs, unfiltered [attribute-error]
 # mercurial/win32.py            # [not-callable]
 # mercurial/wireprotoframing.py # [unsupported-operands], [attribute-error], [import-error]
@@ -43,7 +72,37 b' cd `hg root`'
 
 # TODO: include hgext and hgext3rd
 
-pytype -V 3.7 --keep-going --jobs auto mercurial \
+pytype -V 3.7 --keep-going --jobs auto \
+    doc/check-seclevel.py hgdemandimport hgext mercurial \
+    -x hgext/absorb.py \
+    -x hgext/bugzilla.py \
+    -x hgext/convert/bzr.py \
+    -x hgext/convert/cvs.py \
+    -x hgext/convert/cvsps.py \
+    -x hgext/convert/p4.py \
+    -x hgext/convert/subversion.py \
+    -x hgext/fastannotate/context.py \
+    -x hgext/fastannotate/formatter.py \
+    -x hgext/fsmonitor/__init__.py \
+    -x hgext/git/__init__.py \
+    -x hgext/githelp.py \
+    -x hgext/hgk.py \
+    -x hgext/histedit.py \
+    -x hgext/infinitepush \
+    -x hgext/keyword.py \
+    -x hgext/largefiles/storefactory.py \
+    -x hgext/lfs/__init__.py \
+    -x hgext/narrow/narrowbundle2.py \
+    -x hgext/narrow/narrowcommands.py \
+    -x hgext/rebase.py \
+    -x hgext/remotefilelog/basepack.py \
+    -x hgext/remotefilelog/basestore.py \
+    -x hgext/remotefilelog/contentstore.py \
+    -x hgext/remotefilelog/fileserverclient.py \
+    -x hgext/remotefilelog/remotefilectx.py \
+    -x hgext/remotefilelog/shallowbundle.py \
+    -x hgext/sqlitestore.py \
+    -x hgext/zeroconf/__init__.py \
     -x mercurial/bundlerepo.py \
     -x mercurial/context.py \
     -x mercurial/crecord.py \
@@ -64,9 +123,11 b' pytype -V 3.7 --keep-going --jobs auto m'
     -x mercurial/repoview.py \
     -x mercurial/testing/storage.py \
     -x mercurial/thirdparty \
-    -x mercurial/ui.py \
     -x mercurial/unionrepo.py \
     -x mercurial/win32.py \
     -x mercurial/wireprotoframing.py \
     -x mercurial/wireprotov1peer.py \
     -x mercurial/wireprotov1server.py
+
+echo 'pytype crashed while generating the following type stubs:'
+find .pytype/pyi -name '*.pyi' | xargs grep -l '# Caught error' | sort

contrib/fuzz/revlog.cc

@@ -20,7 +20,7 b' for inline in (True, False):'
         index, cache = parsers.parse_index2(data, inline)
         index.slicechunktodensity(list(range(len(index))), 0.5, 262144)
         index.stats()
-        index.findsnapshots({}, 0)
+        index.findsnapshots({}, 0, len(index) - 1)
         10 in index
         for rev in range(len(index)):
             index.reachableroots(0, [len(index)-1], [rev])

contrib/heptapod-ci.yml

@@ -42,6 +42,7 b' rust-cargo-test:'
     script:
         - echo "python used, $PYTHON"
         - make rust-tests
+        - make cargo-clippy
     variables:
         PYTHON: python3
         CI_CLEVER_CLOUD_FLAVOR: S
@@ -91,7 +92,8 b' check-pytype:'
       - hg -R /tmp/mercurial-ci/ update `hg log --rev '.' --template '{node}'`
       - cd /tmp/mercurial-ci/
       - make local PYTHON=$PYTHON
-      - $PYTHON -m pip install --user -U libcst==0.3.20 pytype==2022.03.29
+      - $PYTHON -m pip install --user -U libcst==0.3.20 pytype==2022.11.18
+      - ./contrib/setup-pytype.sh
     script:
       - echo "Entering script section"
       - sh contrib/check-pytype.sh

contrib/perf.py

@@ -235,6 +235,7 b' revlogopts = getattr('
 
 cmdtable = {}
 
+
 # for "historical portability":
 # define parsealiases locally, because cmdutil.parsealiases has been
 # available since 1.5 (or 6252852b4332)
@@ -573,7 +574,6 b' def _timer('
 
 
 def formatone(fm, timings, title=None, result=None, displayall=False):
-
     count = len(timings)
 
     fm.startitem()
@@ -815,7 +815,12 b' def perfstatus(ui, repo, **opts):'
             )
             sum(map(bool, s))
 
-        timer(status_dirstate)
+        if util.safehasattr(dirstate, 'running_status'):
+            with dirstate.running_status(repo):
+                timer(status_dirstate)
+                dirstate.invalidate()
+        else:
+            timer(status_dirstate)
     else:
         timer(lambda: sum(map(len, repo.status(unknown=opts[b'unknown']))))
     fm.end()
@@ -997,11 +1002,16 b' def perfdiscovery(ui, repo, path, **opts'
     timer, fm = gettimer(ui, opts)
 
     try:
-        from mercurial.utils.urlutil import get_unique_pull_path
-
-        path = get_unique_pull_path(b'perfdiscovery', repo, ui, path)[0]
+        from mercurial.utils.urlutil import get_unique_pull_path_obj
+
+        path = get_unique_pull_path_obj(b'perfdiscovery', ui, path)
     except ImportError:
-        path = ui.expandpath(path)
+        try:
+            from mercurial.utils.urlutil import get_unique_pull_path
+
+            path = get_unique_pull_path(b'perfdiscovery', repo, ui, path)[0]
+        except ImportError:
+            path = ui.expandpath(path)
 
     def s():
         repos[1] = hg.peer(ui, opts, path)
@@ -1469,7 +1479,8 b' def perfdirstatewrite(ui, repo, **opts):'
     def d():
         ds.write(repo.currenttransaction())
 
-    timer(d, setup=setup)
+    with repo.wlock():
+        timer(d, setup=setup)
     fm.end()
 
 
@@ -1613,7 +1624,11 b' def perfphasesremote(ui, repo, dest=None'
             b'default repository not configured!',
             hint=b"see 'hg help config.paths'",
         )
-    dest = path.pushloc or path.loc
+    if util.safehasattr(path, 'main_path'):
+        path = path.get_push_variant()
+        dest = path.loc
+    else:
+        dest = path.pushloc or path.loc
     ui.statusnoi18n(b'analysing phase of %s\n' % util.hidepassword(dest))
     other = hg.peer(repo, opts, dest)
 

contrib/testparseutil.py

@@ -7,14 +7,12 b''
 
 
 import abc
+import builtins
 import re
-import sys
 
 ####################
 # for Python3 compatibility (almost comes from mercurial/pycompat.py)
 
-ispy3 = sys.version_info[0] >= 3
-
 
 def identity(a):
     return a
@@ -38,27 +36,19 b' def rapply(f, xs):'
     return _rapply(f, xs)
 
 
-if ispy3:
-    import builtins
-
-    def bytestr(s):
-        # tiny version of pycompat.bytestr
-        return s.encode('latin1')
-
-    def sysstr(s):
-        if isinstance(s, builtins.str):
-            return s
-        return s.decode('latin-1')
-
-    def opentext(f):
-        return open(f, 'r')
+def bytestr(s):
+    # tiny version of pycompat.bytestr
+    return s.encode('latin1')
 
 
-else:
-    bytestr = str
-    sysstr = identity
+def sysstr(s):
+    if isinstance(s, builtins.str):
+        return s
+    return s.decode('latin-1')
 
-    opentext = open
+
+def opentext(f):
+    return open(f, 'r')
 
 
 def b2s(x):

doc/check-seclevel.py

@@ -46,7 +46,7 b' def showavailables(ui, initlevel):'
 
 
 def checkseclevel(ui, doc, name, initlevel):
-    ui.notenoi18n('checking "%s"\n' % name)
+    ui.notenoi18n(('checking "%s"\n' % name).encode('utf-8'))
     if not isinstance(doc, bytes):
         doc = doc.encode('utf-8')
     blocks, pruned = minirst.parse(doc, 0, ['verbose'])
@@ -70,14 +70,18 b' def checkseclevel(ui, doc, name, initlev'
         nextlevel = mark2level[mark]
         if curlevel < nextlevel and curlevel + 1 != nextlevel:
             ui.warnnoi18n(
-                'gap of section level at "%s" of %s\n' % (title, name)
+                ('gap of section level at "%s" of %s\n' % (title, name)).encode(
+                    'utf-8'
+                )
             )
             showavailables(ui, initlevel)
             errorcnt += 1
             continue
         ui.notenoi18n(
-            'appropriate section level for "%s %s"\n'
-            % (mark * (nextlevel * 2), title)
+            (
+                'appropriate section level for "%s %s"\n'
+                % (mark * (nextlevel * 2), title)
+            ).encode('utf-8')
         )
         curlevel = nextlevel
 
@@ -90,7 +94,9 b' def checkcmdtable(ui, cmdtable, namefmt,'
         name = k.split(b"|")[0].lstrip(b"^")
         if not entry[0].__doc__:
             ui.notenoi18n(
-                'skip checking %s: no help document\n' % (namefmt % name)
+                (
+                    'skip checking %s: no help document\n' % (namefmt % name)
+                ).encode('utf-8')
             )
             continue
         errorcnt += checkseclevel(
@@ -117,7 +123,9 b' def checkhghelps(ui):'
         mod = extensions.load(ui, name, None)
         if not mod.__doc__:
             ui.notenoi18n(
-                'skip checking %s extension: no help document\n' % name
+                (
+                    'skip checking %s extension: no help document\n' % name
+                ).encode('utf-8')
             )
             continue
         errorcnt += checkseclevel(
@@ -144,7 +152,9 b' def checkfile(ui, filename, initlevel):'
             doc = fp.read()
 
     ui.notenoi18n(
-        'checking input from %s with initlevel %d\n' % (filename, initlevel)
+        (
+            'checking input from %s with initlevel %d\n' % (filename, initlevel)
+        ).encode('utf-8')
     )
     return checkseclevel(ui, doc, 'input from %s' % filename, initlevel)
 

hgdemandimport/demandimportpy3.py

@@ -23,8 +23,6 b' This also has some limitations compared '
   enabled.
 """
 
-# This line is unnecessary, but it satisfies test-check-py3-compat.t.
-
 import contextlib
 import importlib.util
 import sys
@@ -39,10 +37,16 b' class _lazyloaderex(importlib.util.LazyL'
     the ignore list.
     """
 
+    _HAS_DYNAMIC_ATTRIBUTES = True  # help pytype not flag self.loader
+
     def exec_module(self, module):
         """Make the module load lazily."""
         with tracing.log('demandimport %s', module):
             if _deactivated or module.__name__ in ignores:
+                # Reset the loader on the module as super() does (issue6725)
+                module.__spec__.loader = self.loader
+                module.__loader__ = self.loader
+
                 self.loader.exec_module(module)
             else:
                 super().exec_module(module)

hgext/absorb.py

@@ -881,7 +881,7 b' class fixupstate:'
 
             dirstate._fsmonitorstate.invalidate = noop
         try:
-            with dirstate.parentchange():
+            with dirstate.changing_parents(self.repo):
                 dirstate.rebuild(ctx.node(), ctx.manifest(), self.paths)
         finally:
             restore()

hgext/amend.py

@@ -46,6 +46,7 b' command = registrar.command(cmdtable)'
             _(b'mark a branch as closed, hiding it from the branch list'),
         ),
         (b's', b'secret', None, _(b'use the secret phase for committing')),
+        (b'', b'draft', None, _(b'use the draft phase for committing')),
         (b'n', b'note', b'', _(b'store a note on the amend')),
     ]
     + cmdutil.walkopts
@@ -64,6 +65,7 b' def amend(ui, repo, *pats, **opts):'
 
     See :hg:`help commit` for more details.
     """
+    cmdutil.check_at_most_one_arg(opts, 'draft', 'secret')
     cmdutil.check_note_size(opts)
 
     with repo.wlock(), repo.lock():

hgext/automv.py

@@ -59,21 +59,29 b' def mvcheck(orig, ui, repo, *pats, **opt'
     opts = pycompat.byteskwargs(opts)
     renames = None
     disabled = opts.pop(b'no_automv', False)
-    if not disabled:
-        threshold = ui.configint(b'automv', b'similarity')
-        if not 0 <= threshold <= 100:
-            raise error.Abort(_(b'automv.similarity must be between 0 and 100'))
-        if threshold > 0:
-            match = scmutil.match(repo[None], pats, opts)
-            added, removed = _interestingfiles(repo, match)
-            uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=True)
-            renames = _findrenames(
-                repo, uipathfn, added, removed, threshold / 100.0
-            )
+    with repo.wlock():
+        if not disabled:
+            threshold = ui.configint(b'automv', b'similarity')
+            if not 0 <= threshold <= 100:
+                raise error.Abort(
+                    _(b'automv.similarity must be between 0 and 100')
+                )
+            if threshold > 0:
+                match = scmutil.match(repo[None], pats, opts)
+                added, removed = _interestingfiles(repo, match)
+                uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=True)
+                renames = _findrenames(
+                    repo, uipathfn, added, removed, threshold / 100.0
+                )
 
-    with repo.wlock():
         if renames is not None:
-            scmutil._markchanges(repo, (), (), renames)
+            with repo.dirstate.changing_files(repo):
+                # XXX this should be wider and integrated with the commit
+                # transaction. At the same time as we do the `addremove` logic
+                # for commit.  However we can't really do better with the
+                # current extension structure, and this is not worse than what
+                # happened before.
+                scmutil._markchanges(repo, (), (), renames)
         return orig(ui, repo, *pats, **pycompat.strkwargs(opts))
 
 

hgext/blackbox.py

@@ -217,6 +217,8 b' def blackbox(ui, repo, *revs, **opts):'
         return
 
     limit = opts.get('limit')
+    assert limit is not None  # help pytype
+
     fp = repo.vfs(b'blackbox.log', b'r')
     lines = fp.read().split(b'\n')
 

hgext/convert/bzr.py

@@ -31,11 +31,14 b' demandimport.IGNORES.update('
 
 try:
     # bazaar imports
+    # pytype: disable=import-error
     import breezy.bzr.bzrdir
     import breezy.errors
     import breezy.revision
     import breezy.revisionspec
 
+    # pytype: enable=import-error
+
     bzrdir = breezy.bzr.bzrdir
     errors = breezy.errors
     revision = breezy.revision

hgext/convert/hg.py

@@ -608,7 +608,10 b' class mercurial_source(common.converter_'
             files = copyfiles = ctx.manifest()
         if parents:
             if self._changescache[0] == rev:
-                ma, r = self._changescache[1]
+                # TODO: add type hints to avoid this warning, instead of
+                #  suppressing it:
+                #     No attribute '__iter__' on None [attribute-error]
+                ma, r = self._changescache[1]  # pytype: disable=attribute-error
             else:
                 ma, r = self._changedfiles(parents[0], ctx)
             if not full:

hgext/convert/monotone.py

@@ -243,6 +243,7 b' class monotone_source(common.converter_s'
             m = self.cert_re.match(e)
             if m:
                 name, value = m.groups()
+                assert value is not None  # help pytype
                 value = value.replace(br'\"', b'"')
                 value = value.replace(br'\\', b'\\')
                 certs[name] = value

hgext/convert/subversion.py

@@ -47,11 +47,14 b' NoRepo = common.NoRepo'
 # these bindings.
 
 try:
+    # pytype: disable=import-error
     import svn
     import svn.client
     import svn.core
     import svn.ra
     import svn.delta
+
+    # pytype: enable=import-error
     from . import transport
     import warnings
 
@@ -722,7 +725,13 b' class svn_source(converter_source):'
     def getchanges(self, rev, full):
         # reuse cache from getchangedfiles
         if self._changescache[0] == rev and not full:
+            # TODO: add type hints to avoid this warning, instead of
+            #  suppressing it:
+            #     No attribute '__iter__' on None [attribute-error]
+
+            # pytype: disable=attribute-error
             (files, copies) = self._changescache[1]
+            # pytype: enable=attribute-error
         else:
             (files, copies) = self._getchanges(rev, full)
             # caller caches the result, so free it here to release memory

hgext/convert/transport.py

@@ -17,10 +17,13 b''
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, see <http://www.gnu.org/licenses/>.
 
+# pytype: disable=import-error
 import svn.client
 import svn.core
 import svn.ra
 
+# pytype: enable=import-error
+
 Pool = svn.core.Pool
 SubversionException = svn.core.SubversionException
 
@@ -37,7 +40,7 b' svn_config = None'
 
 def _create_auth_baton(pool):
     """Create a Subversion authentication baton."""
-    import svn.client
+    import svn.client  # pytype: disable=import-error
 
     # Give the client context baton a suite of authentication
     # providers.h

hgext/eol.py

@@ -421,30 +421,31 b' def reposetup(ui, repo):'
                 wlock = None
                 try:
                     wlock = self.wlock()
-                    for f in self.dirstate:
-                        if not self.dirstate.get_entry(f).maybe_clean:
-                            continue
-                        if oldeol is not None:
-                            if not oldeol.match(f) and not neweol.match(f):
+                    with self.dirstate.changing_files(self):
+                        for f in self.dirstate:
+                            if not self.dirstate.get_entry(f).maybe_clean:
                                 continue
-                            oldkey = None
-                            for pattern, key, m in oldeol.patterns:
-                                if m(f):
-                                    oldkey = key
-                                    break
-                            newkey = None
-                            for pattern, key, m in neweol.patterns:
-                                if m(f):
-                                    newkey = key
-                                    break
-                            if oldkey == newkey:
-                                continue
-                        # all normal files need to be looked at again since
-                        # the new .hgeol file specify a different filter
-                        self.dirstate.set_possibly_dirty(f)
-                    # Write the cache to update mtime and cache .hgeol
-                    with self.vfs(b"eol.cache", b"w") as f:
-                        f.write(hgeoldata)
+                            if oldeol is not None:
+                                if not oldeol.match(f) and not neweol.match(f):
+                                    continue
+                                oldkey = None
+                                for pattern, key, m in oldeol.patterns:
+                                    if m(f):
+                                        oldkey = key
+                                        break
+                                newkey = None
+                                for pattern, key, m in neweol.patterns:
+                                    if m(f):
+                                        newkey = key
+                                        break
+                                if oldkey == newkey:
+                                    continue
+                            # all normal files need to be looked at again since
+                            # the new .hgeol file specify a different filter
+                            self.dirstate.set_possibly_dirty(f)
+                        # Write the cache to update mtime and cache .hgeol
+                        with self.vfs(b"eol.cache", b"w") as f:
+                            f.write(hgeoldata)
                 except errormod.LockUnavailable:
                     # If we cannot lock the repository and clear the
                     # dirstate, then a commit might not see all files

hgext/fastannotate/protocol.py

@@ -151,8 +151,11 b' def annotatepeer(repo):'
     ui = repo.ui
 
     remotedest = ui.config(b'fastannotate', b'remotepath', b'default')
-    r = urlutil.get_unique_pull_path(b'fastannotate', repo, ui, remotedest)
-    remotepath = r[0]
+    remotepath = urlutil.get_unique_pull_path_obj(
+        b'fastannotate',
+        ui,
+        remotedest,
+    )
     peer = hg.peer(ui, {}, remotepath)
 
     try:

hgext/fetch.py

@@ -108,9 +108,9 b" def fetch(ui, repo, source=b'default', *"
                 )
             )
 
-        path = urlutil.get_unique_pull_path(b'fetch', repo, ui, source)[0]
+        path = urlutil.get_unique_pull_path_obj(b'fetch', ui, source)
         other = hg.peer(repo, opts, path)
-        ui.status(_(b'pulling from %s\n') % urlutil.hidepassword(path))
+        ui.status(_(b'pulling from %s\n') % urlutil.hidepassword(path.loc))
         revs = None
         if opts[b'rev']:
             try:

hgext/fix.py

@@ -779,7 +779,7 b' def writeworkingdir(repo, ctx, filedata,'
     newp1 = replacements.get(oldp1, oldp1)
     if newp1 != oldp1:
         assert repo.dirstate.p2() == nullid
-        with repo.dirstate.parentchange():
+        with repo.dirstate.changing_parents(repo):
             scmutil.movedirstate(repo, repo[newp1])
 
 

hgext/fsmonitor/pywatchman/__init__.py

@@ -26,8 +26,6 b''
 # OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
-# no unicode literals
-
 import inspect
 import math
 import os
@@ -94,7 +92,9 b' if os.name == "nt":'
 
     LPDWORD = ctypes.POINTER(wintypes.DWORD)
 
-    CreateFile = ctypes.windll.kernel32.CreateFileA
+    _kernel32 = ctypes.windll.kernel32  # pytype: disable=module-attr
+
+    CreateFile = _kernel32.CreateFileA
     CreateFile.argtypes = [
         wintypes.LPSTR,
         wintypes.DWORD,
@@ -106,11 +106,11 b' if os.name == "nt":'
     ]
     CreateFile.restype = wintypes.HANDLE
 
-    CloseHandle = ctypes.windll.kernel32.CloseHandle
+    CloseHandle = _kernel32.CloseHandle
     CloseHandle.argtypes = [wintypes.HANDLE]
     CloseHandle.restype = wintypes.BOOL
 
-    ReadFile = ctypes.windll.kernel32.ReadFile
+    ReadFile = _kernel32.ReadFile
     ReadFile.argtypes = [
         wintypes.HANDLE,
         wintypes.LPVOID,
@@ -120,7 +120,7 b' if os.name == "nt":'
     ]
     ReadFile.restype = wintypes.BOOL
 
-    WriteFile = ctypes.windll.kernel32.WriteFile
+    WriteFile = _kernel32.WriteFile
     WriteFile.argtypes = [
         wintypes.HANDLE,
         wintypes.LPVOID,
@@ -130,15 +130,15 b' if os.name == "nt":'
     ]
     WriteFile.restype = wintypes.BOOL
 
-    GetLastError = ctypes.windll.kernel32.GetLastError
+    GetLastError = _kernel32.GetLastError
     GetLastError.argtypes = []
     GetLastError.restype = wintypes.DWORD
 
-    SetLastError = ctypes.windll.kernel32.SetLastError
+    SetLastError = _kernel32.SetLastError
     SetLastError.argtypes = [wintypes.DWORD]
     SetLastError.restype = None
 
-    FormatMessage = ctypes.windll.kernel32.FormatMessageA
+    FormatMessage = _kernel32.FormatMessageA
     FormatMessage.argtypes = [
         wintypes.DWORD,
         wintypes.LPVOID,
@@ -150,9 +150,9 b' if os.name == "nt":'
     ]
     FormatMessage.restype = wintypes.DWORD
 
-    LocalFree = ctypes.windll.kernel32.LocalFree
+    LocalFree = _kernel32.LocalFree
 
-    GetOverlappedResult = ctypes.windll.kernel32.GetOverlappedResult
+    GetOverlappedResult = _kernel32.GetOverlappedResult
     GetOverlappedResult.argtypes = [
         wintypes.HANDLE,
         ctypes.POINTER(OVERLAPPED),
@@ -161,9 +161,7 b' if os.name == "nt":'
     ]
     GetOverlappedResult.restype = wintypes.BOOL
 
-    GetOverlappedResultEx = getattr(
-        ctypes.windll.kernel32, "GetOverlappedResultEx", None
-    )
+    GetOverlappedResultEx = getattr(_kernel32, "GetOverlappedResultEx", None)
     if GetOverlappedResultEx is not None:
         GetOverlappedResultEx.argtypes = [
             wintypes.HANDLE,
@@ -174,7 +172,7 b' if os.name == "nt":'
         ]
         GetOverlappedResultEx.restype = wintypes.BOOL
 
-    WaitForSingleObjectEx = ctypes.windll.kernel32.WaitForSingleObjectEx
+    WaitForSingleObjectEx = _kernel32.WaitForSingleObjectEx
     WaitForSingleObjectEx.argtypes = [
         wintypes.HANDLE,
         wintypes.DWORD,
@@ -182,7 +180,7 b' if os.name == "nt":'
     ]
     WaitForSingleObjectEx.restype = wintypes.DWORD
 
-    CreateEvent = ctypes.windll.kernel32.CreateEventA
+    CreateEvent = _kernel32.CreateEventA
     CreateEvent.argtypes = [
         LPDWORD,
         wintypes.BOOL,
@@ -192,7 +190,7 b' if os.name == "nt":'
     CreateEvent.restype = wintypes.HANDLE
 
     # Windows Vista is the minimum supported client for CancelIoEx.
-    CancelIoEx = ctypes.windll.kernel32.CancelIoEx
+    CancelIoEx = _kernel32.CancelIoEx
     CancelIoEx.argtypes = [wintypes.HANDLE, ctypes.POINTER(OVERLAPPED)]
     CancelIoEx.restype = wintypes.BOOL
 
@@ -691,9 +689,9 b' class CLIProcessTransport(Transport):'
         if self.closed:
             self.close()
             self.closed = False
-        self._connect()
-        res = self.proc.stdin.write(data)
-        self.proc.stdin.close()
+        proc = self._connect()
+        res = proc.stdin.write(data)
+        proc.stdin.close()
         self.closed = True
         return res
 
@@ -988,8 +986,12 b' class client:'
                 # if invoked via an application with graphical user interface,
                 # this call will cause a brief command window pop-up.
                 # Using the flag STARTF_USESHOWWINDOW to avoid this behavior.
+
+                # pytype: disable=module-attr
                 startupinfo = subprocess.STARTUPINFO()
                 startupinfo.dwFlags |= subprocess.STARTF_USESHOWWINDOW
+                # pytype: enable=module-attr
+
                 args["startupinfo"] = startupinfo
 
             p = subprocess.Popen(cmd, **args)
@@ -1026,7 +1028,11 b' class client:'
         if self.transport == CLIProcessTransport:
             kwargs["binpath"] = self.binpath
 
+        # Only CLIProcessTransport has the binpath kwarg
+        # pytype: disable=wrong-keyword-args
         self.tport = self.transport(self.sockpath, self.timeout, **kwargs)
+        # pytype: enable=wrong-keyword-args
+
         self.sendConn = self.sendCodec(self.tport)
         self.recvConn = self.recvCodec(self.tport)
         self.pid = os.getpid()

hgext/fsmonitor/pywatchman/capabilities.py

@@ -26,8 +26,6 b''
 # OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
-# no unicode literals
-
 
 def parse_version(vstr):
     res = 0

hgext/fsmonitor/pywatchman/compat.py

@@ -26,45 +26,28 b''
 # OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
-# no unicode literals
-
 import sys
 
 
 """Compatibility module across Python 2 and 3."""
 
 
-PYTHON2 = sys.version_info < (3, 0)
 PYTHON3 = sys.version_info >= (3, 0)
 
 # This is adapted from https://bitbucket.org/gutworth/six, and used under the
 # MIT license. See LICENSE for a full copyright notice.
-if PYTHON3:
-
-    def reraise(tp, value, tb=None):
-        try:
-            if value is None:
-                value = tp()
-            if value.__traceback__ is not tb:
-                raise value.with_traceback(tb)
-            raise value
-        finally:
-            value = None
-            tb = None
 
 
-else:
-    exec(
-        """
 def reraise(tp, value, tb=None):
     try:
-        raise tp, value, tb
+        if value is None:
+            value = tp()
+        if value.__traceback__ is not tb:
+            raise value.with_traceback(tb)
+        raise value
     finally:
+        value = None
         tb = None
-""".strip()
-    )
+
 
-if PYTHON3:
-    UNICODE = str
-else:
-    UNICODE = unicode  # noqa: F821 We handled versioning above
+UNICODE = str

hgext/fsmonitor/pywatchman/encoding.py

@@ -26,8 +26,6 b''
 # OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
-# no unicode literals
-
 import sys
 
 from . import compat

hgext/fsmonitor/pywatchman/load.py

@@ -26,8 +26,6 b''
 # OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
-# no unicode literals
-
 import ctypes
 
 

hgext/fsmonitor/pywatchman/pybser.py

@@ -26,8 +26,6 b''
 # OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
-# no unicode literals
-
 import binascii
 import collections
 import ctypes
@@ -53,17 +51,15 b' BSER_TEMPLATE = b"\\x0b"'
 BSER_SKIP = b"\x0c"
 BSER_UTF8STRING = b"\x0d"
 
-if compat.PYTHON3:
-    STRING_TYPES = (str, bytes)
-    unicode = str
+STRING_TYPES = (str, bytes)
+unicode = str
+
 
-    def tobytes(i):
-        return str(i).encode("ascii")
+def tobytes(i):
+    return str(i).encode("ascii")
 
-    long = int
-else:
-    STRING_TYPES = (unicode, str)
-    tobytes = bytes
+
+long = int
 
 # Leave room for the serialization header, which includes
 # our overall length.  To make things simpler, we'll use an
@@ -89,7 +85,7 b' def _int_size(x):'
 def _buf_pos(buf, pos):
     ret = buf[pos]
     # Normalize the return type to bytes
-    if compat.PYTHON3 and not isinstance(ret, bytes):
+    if not isinstance(ret, bytes):
         ret = bytes((ret,))
     return ret
 
@@ -252,10 +248,7 b' class _bser_buffer:'
             else:
                 raise RuntimeError("Cannot represent this mapping value")
             self.wpos += needed
-            if compat.PYTHON3:
-                iteritems = val.items()
-            else:
-                iteritems = val.iteritems()  # noqa: B301 Checked version above
+            iteritems = val.items()
             for k, v in iteritems:
                 self.append_string(k)
                 self.append_recursive(v)

hgext/git/dirstate.py

@@ -260,7 +260,12 b' class gitdirstate:'
     # # TODO what the heck is this
     _filecache = set()
 
-    def pendingparentchange(self):
+    def is_changing_parents(self):
+        # TODO: we need to implement the context manager bits and
+        # correctly stage/revert index edits.
+        return False
+
+    def is_changing_any(self):
         # TODO: we need to implement the context manager bits and
         # correctly stage/revert index edits.
         return False
@@ -322,14 +327,6 b' class gitdirstate:'
             r[path] = s
         return r
 
-    def savebackup(self, tr, backupname):
-        # TODO: figure out a strategy for saving index backups.
-        pass
-
-    def restorebackup(self, tr, backupname):
-        # TODO: figure out a strategy for saving index backups.
-        pass
-
     def set_tracked(self, f, reset_copy=False):
         # TODO: support copies and reset_copy=True
         uf = pycompat.fsdecode(f)
@@ -384,7 +381,7 b' class gitdirstate:'
         pass
 
     @contextlib.contextmanager
-    def parentchange(self):
+    def changing_parents(self, repo):
         # TODO: track this maybe?
         yield
 
@@ -392,10 +389,6 b' class gitdirstate:'
         # TODO: should this be added to the dirstate interface?
         self._plchangecallbacks[category] = callback
 
-    def clearbackup(self, tr, backupname):
-        # TODO
-        pass
-
     def setbranch(self, branch):
         raise error.Abort(
             b'git repos do not support branches. try using bookmarks'

hgext/git/gitutil.py

@@ -9,7 +9,7 b' def get_pygit2():'
     global pygit2_module
     if pygit2_module is None:
         try:
-            import pygit2 as pygit2_module
+            import pygit2 as pygit2_module  # pytype: disable=import-error
 
             pygit2_module.InvalidSpecError
         except (ImportError, AttributeError):

hgext/gpg.py

@@ -352,7 +352,8 b' def _dosign(ui, repo, *revs, **opts):'
     sigsfile.close()
 
     if b'.hgsigs' not in repo.dirstate:
-        repo[None].add([b".hgsigs"])
+        with repo.dirstate.changing_files(repo):
+            repo[None].add([b".hgsigs"])
 
     if opts[b"no_commit"]:
         return

hgext/histedit.py

@@ -1051,12 +1051,11 b' def findoutgoing(ui, repo, remote=None, '
     if opts is None:
         opts = {}
     path = urlutil.get_unique_push_path(b'histedit', repo, ui, remote)
-    dest = path.pushloc or path.loc
-
-    ui.status(_(b'comparing with %s\n') % urlutil.hidepassword(dest))
+
+    ui.status(_(b'comparing with %s\n') % urlutil.hidepassword(path.loc))
 
     revs, checkout = hg.addbranchrevs(repo, repo, (path.branch, []), None)
-    other = hg.peer(repo, opts, dest)
+    other = hg.peer(repo, opts, path)
 
     if revs:
         revs = [repo.lookup(rev) for rev in revs]

hgext/hooklib/changeset_obsoleted.py

@@ -32,7 +32,10 b' from mercurial import ('
     pycompat,
     registrar,
 )
-from mercurial.utils import dateutil
+from mercurial.utils import (
+    dateutil,
+    stringutil,
+)
 from .. import notify
 
 configtable = {}
@@ -98,7 +101,7 b' def _report_commit(ui, repo, ctx):'
     try:
         msg = mail.parsebytes(data)
     except emailerrors.MessageParseError as inst:
-        raise error.Abort(inst)
+        raise error.Abort(stringutil.forcebytestr(inst))
 
     msg['In-reply-to'] = notify.messageid(ctx, domain, messageidseed)
     msg['Message-Id'] = notify.messageid(

hgext/hooklib/changeset_published.py

@@ -31,7 +31,10 b' from mercurial import ('
     pycompat,
     registrar,
 )
-from mercurial.utils import dateutil
+from mercurial.utils import (
+    dateutil,
+    stringutil,
+)
 from .. import notify
 
 configtable = {}
@@ -97,7 +100,7 b' def _report_commit(ui, repo, ctx):'
     try:
         msg = mail.parsebytes(data)
     except emailerrors.MessageParseError as inst:
-        raise error.Abort(inst)
+        raise error.Abort(stringutil.forcebytestr(inst))
 
     msg['In-reply-to'] = notify.messageid(ctx, domain, messageidseed)
     msg['Message-Id'] = notify.messageid(

hgext/infinitepush/__init__.py

@@ -683,12 +683,10 b' def _lookupwrap(orig):'
 def _pull(orig, ui, repo, source=b"default", **opts):
     opts = pycompat.byteskwargs(opts)
     # Copy paste from `pull` command
-    source, branches = urlutil.get_unique_pull_path(
+    path = urlutil.get_unique_pull_path_obj(
         b"infinite-push's pull",
-        repo,
         ui,
         source,
-        default_branches=opts.get(b'branch'),
     )
 
     scratchbookmarks = {}
@@ -709,7 +707,7 b' def _pull(orig, ui, repo, source=b"defau'
                 bookmarks.append(bookmark)
 
         if scratchbookmarks:
-            other = hg.peer(repo, opts, source)
+            other = hg.peer(repo, opts, path)
             try:
                 fetchedbookmarks = other.listkeyspatterns(
                     b'bookmarks', patterns=scratchbookmarks
@@ -734,14 +732,14 b' def _pull(orig, ui, repo, source=b"defau'
     try:
         # Remote scratch bookmarks will be deleted because remotenames doesn't
         # know about them. Let's save it before pull and restore after
-        remotescratchbookmarks = _readscratchremotebookmarks(ui, repo, source)
-        result = orig(ui, repo, source, **pycompat.strkwargs(opts))
+        remotescratchbookmarks = _readscratchremotebookmarks(ui, repo, path.loc)
+        result = orig(ui, repo, path.loc, **pycompat.strkwargs(opts))
         # TODO(stash): race condition is possible
         # if scratch bookmarks was updated right after orig.
         # But that's unlikely and shouldn't be harmful.
         if common.isremotebooksenabled(ui):
             remotescratchbookmarks.update(scratchbookmarks)
-            _saveremotebookmarks(repo, remotescratchbookmarks, source)
+            _saveremotebookmarks(repo, remotescratchbookmarks, path.loc)
         else:
             _savelocalbookmarks(repo, scratchbookmarks)
         return result
@@ -849,14 +847,14 b' def _push(orig, ui, repo, *dests, **opts'
             raise error.Abort(msg)
 
         path = paths[0]
-        destpath = path.pushloc or path.loc
+        destpath = path.loc
         # Remote scratch bookmarks will be deleted because remotenames doesn't
         # know about them. Let's save it before push and restore after
         remotescratchbookmarks = _readscratchremotebookmarks(ui, repo, destpath)
         result = orig(ui, repo, *dests, **pycompat.strkwargs(opts))
         if common.isremotebooksenabled(ui):
             if bookmark and scratchpush:
-                other = hg.peer(repo, opts, destpath)
+                other = hg.peer(repo, opts, path)
                 try:
                     fetchedbookmarks = other.listkeyspatterns(
                         b'bookmarks', patterns=[bookmark]

hgext/journal.py

@@ -567,8 +567,12 b' def journal(ui, repo, *args, **opts):'
         )
         fm.write(b'newnodes', b'%s', formatnodes(entry.newhashes))
         fm.condwrite(ui.verbose, b'user', b' %-8s', entry.user)
+
+        # ``name`` is bytes, or None only if 'all' was an option.
         fm.condwrite(
+            # pytype: disable=attribute-error
             opts.get(b'all') or name.startswith(b're:'),
+            # pytype: enable=attribute-error
             b'name',
             b'  %-8s',
             entry.name,

hgext/keyword.py

@@ -437,7 +437,7 b' def _kwfwrite(ui, repo, expand, *pats, *'
     if len(wctx.parents()) > 1:
         raise error.Abort(_(b'outstanding uncommitted merge'))
     kwt = getattr(repo, '_keywordkwt', None)
-    with repo.wlock():
+    with repo.wlock(), repo.dirstate.changing_files(repo):
         status = _status(ui, repo, wctx, kwt, *pats, **opts)
         if status.modified or status.added or status.removed or status.deleted:
             raise error.Abort(_(b'outstanding uncommitted changes'))
@@ -530,17 +530,18 b' def demo(ui, repo, *args, **opts):'
     demoitems(b'keywordmaps', kwmaps.items())
     keywords = b'$' + b'$\n$'.join(sorted(kwmaps.keys())) + b'$\n'
     repo.wvfs.write(fn, keywords)
-    repo[None].add([fn])
-    ui.note(_(b'\nkeywords written to %s:\n') % fn)
-    ui.note(keywords)
     with repo.wlock():
+        with repo.dirstate.changing_files(repo):
+            repo[None].add([fn])
+        ui.note(_(b'\nkeywords written to %s:\n') % fn)
+        ui.note(keywords)
         repo.dirstate.setbranch(b'demobranch')
-    for name, cmd in ui.configitems(b'hooks'):
-        if name.split(b'.', 1)[0].find(b'commit') > -1:
-            repo.ui.setconfig(b'hooks', name, b'', b'keyword')
-    msg = _(b'hg keyword configuration and expansion example')
-    ui.note((b"hg ci -m '%s'\n" % msg))
-    repo.commit(text=msg)
+        for name, cmd in ui.configitems(b'hooks'):
+            if name.split(b'.', 1)[0].find(b'commit') > -1:
+                repo.ui.setconfig(b'hooks', name, b'', b'keyword')
+        msg = _(b'hg keyword configuration and expansion example')
+        ui.note((b"hg ci -m '%s'\n" % msg))
+        repo.commit(text=msg)
     ui.status(_(b'\n\tkeywords expanded\n'))
     ui.write(repo.wread(fn))
     repo.wvfs.rmtree(repo.root)
@@ -696,7 +697,7 b' def kw_amend(orig, ui, repo, old, extra,'
     kwt = getattr(repo, '_keywordkwt', None)
     if kwt is None:
         return orig(ui, repo, old, extra, pats, opts)
-    with repo.wlock(), repo.dirstate.parentchange():
+    with repo.wlock(), repo.dirstate.changing_parents(repo):
         kwt.postcommit = True
         newid = orig(ui, repo, old, extra, pats, opts)
         if newid != old.node():
@@ -762,7 +763,7 b' def kw_dorecord(orig, ui, repo, commitfu'
         if ctx != recctx:
             modified, added = _preselect(wstatus, recctx.files())
             kwt.restrict = False
-            with repo.dirstate.parentchange():
+            with repo.dirstate.changing_parents(repo):
                 kwt.overwrite(recctx, modified, False, True)
                 kwt.overwrite(recctx, added, False, True, True)
             kwt.restrict = True

hgext/largefiles/__init__.py

@@ -107,6 +107,7 b' command.'
 
 from mercurial import (
     cmdutil,
+    configitems,
     extensions,
     exthelper,
     hg,
@@ -135,7 +136,7 b' eh.merge(proto.eh)'
 eh.configitem(
     b'largefiles',
     b'minsize',
-    default=eh.configitem.dynamicdefault,
+    default=configitems.dynamicdefault,
 )
 eh.configitem(
     b'largefiles',

hgext/largefiles/lfcommands.py

@@ -219,7 +219,9 b' def lfconvert(ui, src, dest, *pats, **op'
         success = True
     finally:
         if tolfile:
-            rdst.dirstate.clear()
+            # XXX is this the right context semantically ?
+            with rdst.dirstate.changing_parents(rdst):
+                rdst.dirstate.clear()
             release(dstlock, dstwlock)
         if not success:
             # we failed, remove the new directory
@@ -517,53 +519,52 b' def updatelfiles('
             filelist = set(filelist)
             lfiles = [f for f in lfiles if f in filelist]
 
-        with lfdirstate.parentchange():
-            update = {}
-            dropped = set()
-            updated, removed = 0, 0
-            wvfs = repo.wvfs
-            wctx = repo[None]
-            for lfile in lfiles:
-                lfileorig = os.path.relpath(
-                    scmutil.backuppath(ui, repo, lfile), start=repo.root
-                )
-                standin = lfutil.standin(lfile)
-                standinorig = os.path.relpath(
-                    scmutil.backuppath(ui, repo, standin), start=repo.root
-                )
-                if wvfs.exists(standin):
-                    if wvfs.exists(standinorig) and wvfs.exists(lfile):
-                        shutil.copyfile(wvfs.join(lfile), wvfs.join(lfileorig))
-                        wvfs.unlinkpath(standinorig)
-                    expecthash = lfutil.readasstandin(wctx[standin])
-                    if expecthash != b'':
-                        if lfile not in wctx:  # not switched to normal file
-                            if repo.dirstate.get_entry(standin).any_tracked:
-                                wvfs.unlinkpath(lfile, ignoremissing=True)
-                            else:
-                                dropped.add(lfile)
+        update = {}
+        dropped = set()
+        updated, removed = 0, 0
+        wvfs = repo.wvfs
+        wctx = repo[None]
+        for lfile in lfiles:
+            lfileorig = os.path.relpath(
+                scmutil.backuppath(ui, repo, lfile), start=repo.root
+            )
+            standin = lfutil.standin(lfile)
+            standinorig = os.path.relpath(
+                scmutil.backuppath(ui, repo, standin), start=repo.root
+            )
+            if wvfs.exists(standin):
+                if wvfs.exists(standinorig) and wvfs.exists(lfile):
+                    shutil.copyfile(wvfs.join(lfile), wvfs.join(lfileorig))
+                    wvfs.unlinkpath(standinorig)
+                expecthash = lfutil.readasstandin(wctx[standin])
+                if expecthash != b'':
+                    if lfile not in wctx:  # not switched to normal file
+                        if repo.dirstate.get_entry(standin).any_tracked:
+                            wvfs.unlinkpath(lfile, ignoremissing=True)
+                        else:
+                            dropped.add(lfile)
 
-                        # use normallookup() to allocate an entry in largefiles
-                        # dirstate to prevent lfilesrepo.status() from reporting
-                        # missing files as removed.
-                        lfdirstate.update_file(
-                            lfile,
-                            p1_tracked=True,
-                            wc_tracked=True,
-                            possibly_dirty=True,
-                        )
-                        update[lfile] = expecthash
-                else:
-                    # Remove lfiles for which the standin is deleted, unless the
-                    # lfile is added to the repository again. This happens when a
-                    # largefile is converted back to a normal file: the standin
-                    # disappears, but a new (normal) file appears as the lfile.
-                    if (
-                        wvfs.exists(lfile)
-                        and repo.dirstate.normalize(lfile) not in wctx
-                    ):
-                        wvfs.unlinkpath(lfile)
-                        removed += 1
+                    # allocate an entry in largefiles dirstate to prevent
+                    # lfilesrepo.status() from reporting missing files as
+                    # removed.
+                    lfdirstate.hacky_extension_update_file(
+                        lfile,
+                        p1_tracked=True,
+                        wc_tracked=True,
+                        possibly_dirty=True,
+                    )
+                    update[lfile] = expecthash
+            else:
+                # Remove lfiles for which the standin is deleted, unless the
+                # lfile is added to the repository again. This happens when a
+                # largefile is converted back to a normal file: the standin
+                # disappears, but a new (normal) file appears as the lfile.
+                if (
+                    wvfs.exists(lfile)
+                    and repo.dirstate.normalize(lfile) not in wctx
+                ):
+                    wvfs.unlinkpath(lfile)
+                    removed += 1
 
         # largefile processing might be slow and be interrupted - be prepared
         lfdirstate.write(repo.currenttransaction())
@@ -580,41 +581,42 b' def updatelfiles('
             statuswriter(_(b'getting changed largefiles\n'))
             cachelfiles(ui, repo, None, lfiles)
 
-        with lfdirstate.parentchange():
-            for lfile in lfiles:
-                update1 = 0
+        for lfile in lfiles:
+            update1 = 0
 
-                expecthash = update.get(lfile)
-                if expecthash:
-                    if not lfutil.copyfromcache(repo, expecthash, lfile):
-                        # failed ... but already removed and set to normallookup
-                        continue
-                    # Synchronize largefile dirstate to the last modified
-                    # time of the file
-                    lfdirstate.update_file(
-                        lfile, p1_tracked=True, wc_tracked=True
-                    )
+            expecthash = update.get(lfile)
+            if expecthash:
+                if not lfutil.copyfromcache(repo, expecthash, lfile):
+                    # failed ... but already removed and set to normallookup
+                    continue
+                # Synchronize largefile dirstate to the last modified
+                # time of the file
+                lfdirstate.hacky_extension_update_file(
+                    lfile,
+                    p1_tracked=True,
+                    wc_tracked=True,
+                )
+                update1 = 1
+
+            # copy the exec mode of largefile standin from the repository's
+            # dirstate to its state in the lfdirstate.
+            standin = lfutil.standin(lfile)
+            if wvfs.exists(standin):
+                # exec is decided by the users permissions using mask 0o100
+                standinexec = wvfs.stat(standin).st_mode & 0o100
+                st = wvfs.stat(lfile)
+                mode = st.st_mode
+                if standinexec != mode & 0o100:
+                    # first remove all X bits, then shift all R bits to X
+                    mode &= ~0o111
+                    if standinexec:
+                        mode |= (mode >> 2) & 0o111 & ~util.umask
+                    wvfs.chmod(lfile, mode)
                     update1 = 1
 
-                # copy the exec mode of largefile standin from the repository's
-                # dirstate to its state in the lfdirstate.
-                standin = lfutil.standin(lfile)
-                if wvfs.exists(standin):
-                    # exec is decided by the users permissions using mask 0o100
-                    standinexec = wvfs.stat(standin).st_mode & 0o100
-                    st = wvfs.stat(lfile)
-                    mode = st.st_mode
-                    if standinexec != mode & 0o100:
-                        # first remove all X bits, then shift all R bits to X
-                        mode &= ~0o111
-                        if standinexec:
-                            mode |= (mode >> 2) & 0o111 & ~util.umask
-                        wvfs.chmod(lfile, mode)
-                        update1 = 1
+            updated += update1
 
-                updated += update1
-
-                lfutil.synclfdirstate(repo, lfdirstate, lfile, normallookup)
+            lfutil.synclfdirstate(repo, lfdirstate, lfile, normallookup)
 
         lfdirstate.write(repo.currenttransaction())
         if lfiles:

hgext/largefiles/lfutil.py

@@ -159,6 +159,9 b' def findfile(repo, hash):'
 
 
 class largefilesdirstate(dirstate.dirstate):
+    _large_file_dirstate = True
+    _tr_key_suffix = b'-large-files'
+
     def __getitem__(self, key):
         return super(largefilesdirstate, self).__getitem__(unixpath(key))
 
@@ -204,7 +207,13 b' def openlfdirstate(ui, repo, create=True'
     """
     Return a dirstate object that tracks largefiles: i.e. its root is
     the repo root, but it is saved in .hg/largefiles/dirstate.
+
+    If a dirstate object already exists and is being used for a 'changing_*'
+    context, it will be returned.
     """
+    sub_dirstate = getattr(repo.dirstate, '_sub_dirstate', None)
+    if sub_dirstate is not None:
+        return sub_dirstate
     vfs = repo.vfs
     lfstoredir = longname
     opener = vfsmod.vfs(vfs.join(lfstoredir))
@@ -223,20 +232,29 b' def openlfdirstate(ui, repo, create=True'
     # it. This ensures that we create it on the first meaningful
     # largefiles operation in a new clone.
     if create and not vfs.exists(vfs.join(lfstoredir, b'dirstate')):
-        matcher = getstandinmatcher(repo)
-        standins = repo.dirstate.walk(
-            matcher, subrepos=[], unknown=False, ignored=False
-        )
+        try:
+            with repo.wlock(wait=False), lfdirstate.changing_files(repo):
+                matcher = getstandinmatcher(repo)
+                standins = repo.dirstate.walk(
+                    matcher, subrepos=[], unknown=False, ignored=False
+                )
+
+                if len(standins) > 0:
+                    vfs.makedirs(lfstoredir)
 
-        if len(standins) > 0:
-            vfs.makedirs(lfstoredir)
-
-        with lfdirstate.parentchange():
-            for standin in standins:
-                lfile = splitstandin(standin)
-                lfdirstate.update_file(
-                    lfile, p1_tracked=True, wc_tracked=True, possibly_dirty=True
-                )
+                for standin in standins:
+                    lfile = splitstandin(standin)
+                    lfdirstate.hacky_extension_update_file(
+                        lfile,
+                        p1_tracked=True,
+                        wc_tracked=True,
+                        possibly_dirty=True,
+                    )
+        except error.LockError:
+            # Assume that whatever was holding the lock was important.
+            # If we were doing something important, we would already have
+            # either the lock or a largefile dirstate.
+            pass
     return lfdirstate
 
 
@@ -565,10 +583,14 b' def getstandinsstate(repo):'
 def synclfdirstate(repo, lfdirstate, lfile, normallookup):
     lfstandin = standin(lfile)
     if lfstandin not in repo.dirstate:
-        lfdirstate.update_file(lfile, p1_tracked=False, wc_tracked=False)
+        lfdirstate.hacky_extension_update_file(
+            lfile,
+            p1_tracked=False,
+            wc_tracked=False,
+        )
     else:
         entry = repo.dirstate.get_entry(lfstandin)
-        lfdirstate.update_file(
+        lfdirstate.hacky_extension_update_file(
             lfile,
             wc_tracked=entry.tracked,
             p1_tracked=entry.p1_tracked,
@@ -580,8 +602,7 b' def synclfdirstate(repo, lfdirstate, lfi'
 def markcommitted(orig, ctx, node):
     repo = ctx.repo()
 
-    lfdirstate = openlfdirstate(repo.ui, repo)
-    with lfdirstate.parentchange():
+    with repo.dirstate.changing_parents(repo):
         orig(node)
 
         # ATTENTION: "ctx.files()" may differ from "repo[node].files()"
@@ -593,11 +614,11 b' def markcommitted(orig, ctx, node):'
         # - have to be marked as "n" after commit, but
         # - aren't listed in "repo[node].files()"
 
+        lfdirstate = openlfdirstate(repo.ui, repo)
         for f in ctx.files():
             lfile = splitstandin(f)
             if lfile is not None:
                 synclfdirstate(repo, lfdirstate, lfile, False)
-    lfdirstate.write(repo.currenttransaction())
 
     # As part of committing, copy all of the largefiles into the cache.
     #
@@ -668,11 +689,16 b' def updatestandinsbymatch(repo, match):'
         # It can cost a lot of time (several seconds)
         # otherwise to update all standins if the largefiles are
         # large.
-        lfdirstate = openlfdirstate(ui, repo)
         dirtymatch = matchmod.always()
-        unsure, s, mtime_boundary = lfdirstate.status(
-            dirtymatch, subrepos=[], ignored=False, clean=False, unknown=False
-        )
+        with repo.dirstate.running_status(repo):
+            lfdirstate = openlfdirstate(ui, repo)
+            unsure, s, mtime_boundary = lfdirstate.status(
+                dirtymatch,
+                subrepos=[],
+                ignored=False,
+                clean=False,
+                unknown=False,
+            )
         modifiedfiles = unsure + s.modified + s.added + s.removed
         lfiles = listlfiles(repo)
         # this only loops through largefiles that exist (not

hgext/largefiles/overrides.py

@@ -8,6 +8,7 b''
 
 '''Overridden Mercurial commands and functions for the largefiles extension'''
 
+import contextlib
 import copy
 import os
 
@@ -21,6 +22,7 b' from mercurial import ('
     archival,
     cmdutil,
     copies as copiesmod,
+    dirstate,
     error,
     exchange,
     extensions,
@@ -311,6 +313,48 b' def cmdutilremove('
     )
 
 
[e-mail](dirstate.dirstate, b'_changing')
[e-mail]
+def _changing(orig, self, repo, change_type):
+    pre = sub_dirstate = getattr(self, '_sub_dirstate', None)
+    try:
+        lfd = getattr(self, '_large_file_dirstate', False)
+        if sub_dirstate is None and not lfd:
+            sub_dirstate = lfutil.openlfdirstate(repo.ui, repo)
+            self._sub_dirstate = sub_dirstate
+        if not lfd:
+            assert self._sub_dirstate is not None
+        with orig(self, repo, change_type):
+            if sub_dirstate is None:
+                yield
+            else:
+                with sub_dirstate._changing(repo, change_type):
+                    yield
+    finally:
+        self._sub_dirstate = pre
+
+
[e-mail](dirstate.dirstate, b'running_status')
[e-mail]
+def running_status(orig, self, repo):
+    pre = sub_dirstate = getattr(self, '_sub_dirstate', None)
+    try:
+        lfd = getattr(self, '_large_file_dirstate', False)
+        if sub_dirstate is None and not lfd:
+            sub_dirstate = lfutil.openlfdirstate(repo.ui, repo)
+            self._sub_dirstate = sub_dirstate
+        if not lfd:
+            assert self._sub_dirstate is not None
+        with orig(self, repo):
+            if sub_dirstate is None:
+                yield
+            else:
+                with sub_dirstate.running_status(repo):
+                    yield
+    finally:
+        self._sub_dirstate = pre
+
+
 @eh.wrapfunction(subrepo.hgsubrepo, b'status')
 def overridestatusfn(orig, repo, rev2, **opts):
     with lfstatus(repo._repo):
@@ -511,10 +555,12 b' def overridedebugstate(orig, ui, repo, *'
 # largefiles. This makes the merge proceed and we can then handle this
 # case further in the overridden calculateupdates function below.
 @eh.wrapfunction(merge, b'_checkunknownfile')
-def overridecheckunknownfile(origfn, repo, wctx, mctx, f, f2=None):
-    if lfutil.standin(repo.dirstate.normalize(f)) in wctx:
+def overridecheckunknownfile(
+    origfn, dirstate, wvfs, dircache, wctx, mctx, f, f2=None
+):
+    if lfutil.standin(dirstate.normalize(f)) in wctx:
         return False
-    return origfn(repo, wctx, mctx, f, f2)
+    return origfn(dirstate, wvfs, dircache, wctx, mctx, f, f2)
 
 
 # The manifest merge handles conflicts on the manifest level. We want
@@ -658,18 +704,12 b' def overridecalculateupdates('
 def mergerecordupdates(orig, repo, actions, branchmerge, getfiledata):
     if MERGE_ACTION_LARGEFILE_MARK_REMOVED in actions:
         lfdirstate = lfutil.openlfdirstate(repo.ui, repo)
-        with lfdirstate.parentchange():
-            for lfile, args, msg in actions[
-                MERGE_ACTION_LARGEFILE_MARK_REMOVED
-            ]:
-                # this should be executed before 'orig', to execute 'remove'
-                # before all other actions
-                repo.dirstate.update_file(
-                    lfile, p1_tracked=True, wc_tracked=False
-                )
-                # make sure lfile doesn't get synclfdirstate'd as normal
-                lfdirstate.update_file(lfile, p1_tracked=False, wc_tracked=True)
-        lfdirstate.write(repo.currenttransaction())
+        for lfile, args, msg in actions[MERGE_ACTION_LARGEFILE_MARK_REMOVED]:
+            # this should be executed before 'orig', to execute 'remove'
+            # before all other actions
+            repo.dirstate.update_file(lfile, p1_tracked=True, wc_tracked=False)
+            # make sure lfile doesn't get synclfdirstate'd as normal
+            lfdirstate.update_file(lfile, p1_tracked=False, wc_tracked=True)
 
     return orig(repo, actions, branchmerge, getfiledata)
 
@@ -901,7 +941,7 b' def overriderevert(orig, ui, repo, ctx, '
     # Because we put the standins in a bad state (by updating them)
     # and then return them to a correct state we need to lock to
     # prevent others from changing them in their incorrect state.
-    with repo.wlock():
+    with repo.wlock(), repo.dirstate.running_status(repo):
         lfdirstate = lfutil.openlfdirstate(ui, repo)
         s = lfutil.lfdirstatestatus(lfdirstate, repo)
         lfdirstate.write(repo.currenttransaction())
@@ -1436,7 +1476,7 b' def outgoinghook(ui, repo, other, opts, '
 
             def addfunc(fn, lfhash):
                 if fn not in toupload:
-                    toupload[fn] = []
+                    toupload[fn] = []  # pytype: disable=unsupported-operands
                 toupload[fn].append(lfhash)
                 lfhashes.add(lfhash)
 
@@ -1520,20 +1560,34 b' def overridesummary(orig, ui, repo, *pat'
 
 
 @eh.wrapfunction(scmutil, b'addremove')
-def scmutiladdremove(orig, repo, matcher, prefix, uipathfn, opts=None):
+def scmutiladdremove(
+    orig,
+    repo,
+    matcher,
+    prefix,
+    uipathfn,
+    opts=None,
+    open_tr=None,
+):
     if opts is None:
         opts = {}
     if not lfutil.islfilesrepo(repo):
-        return orig(repo, matcher, prefix, uipathfn, opts)
+        return orig(repo, matcher, prefix, uipathfn, opts, open_tr=open_tr)
+
+    # open the transaction and changing_files context
+    if open_tr is not None:
+        open_tr()
+
     # Get the list of missing largefiles so we can remove them
-    lfdirstate = lfutil.openlfdirstate(repo.ui, repo)
-    unsure, s, mtime_boundary = lfdirstate.status(
-        matchmod.always(),
-        subrepos=[],
-        ignored=False,
-        clean=False,
-        unknown=False,
-    )
+    with repo.dirstate.running_status(repo):
+        lfdirstate = lfutil.openlfdirstate(repo.ui, repo)
+        unsure, s, mtime_boundary = lfdirstate.status(
+            matchmod.always(),
+            subrepos=[],
+            ignored=False,
+            clean=False,
+            unknown=False,
+        )
 
     # Call into the normal remove code, but the removing of the standin, we want
     # to have handled by original addremove.  Monkey patching here makes sure
@@ -1567,7 +1621,8 b' def scmutiladdremove(orig, repo, matcher'
     # function to take care of the rest.  Make sure it doesn't do anything with
     # largefiles by passing a matcher that will ignore them.
     matcher = composenormalfilematcher(matcher, repo[None].manifest(), added)
-    return orig(repo, matcher, prefix, uipathfn, opts)
+
+    return orig(repo, matcher, prefix, uipathfn, opts, open_tr=open_tr)
 
 
 # Calling purge with --all will cause the largefiles to be deleted.
@@ -1737,7 +1792,7 b' def mergeupdate(orig, repo, node, branch'
     matcher = kwargs.get('matcher', None)
     # note if this is a partial update
     partial = matcher and not matcher.always()
-    with repo.wlock():
+    with repo.wlock(), repo.dirstate.changing_parents(repo):
         # branch |       |         |
         #  merge | force | partial | action
         # -------+-------+---------+--------------
@@ -1752,15 +1807,15 b' def mergeupdate(orig, repo, node, branch'
         #
         # (*) don't care
         # (*1) deprecated, but used internally (e.g: "rebase --collapse")
-
-        lfdirstate = lfutil.openlfdirstate(repo.ui, repo)
-        unsure, s, mtime_boundary = lfdirstate.status(
-            matchmod.always(),
-            subrepos=[],
-            ignored=False,
-            clean=True,
-            unknown=False,
-        )
+        with repo.dirstate.running_status(repo):
+            lfdirstate = lfutil.openlfdirstate(repo.ui, repo)
+            unsure, s, mtime_boundary = lfdirstate.status(
+                matchmod.always(),
+                subrepos=[],
+                ignored=False,
+                clean=True,
+                unknown=False,
+            )
         oldclean = set(s.clean)
         pctx = repo[b'.']
         dctx = repo[node]
@@ -1787,7 +1842,14 b' def mergeupdate(orig, repo, node, branch'
         # mark all clean largefiles as dirty, just in case the update gets
         # interrupted before largefiles and lfdirstate are synchronized
         for lfile in oldclean:
-            lfdirstate.set_possibly_dirty(lfile)
+            entry = lfdirstate.get_entry(lfile)
+            lfdirstate.hacky_extension_update_file(
+                lfile,
+                wc_tracked=entry.tracked,
+                p1_tracked=entry.p1_tracked,
+                p2_info=entry.p2_info,
+                possibly_dirty=True,
+            )
         lfdirstate.write(repo.currenttransaction())
 
         oldstandins = lfutil.getstandinsstate(repo)
@@ -1798,24 +1860,22 b' def mergeupdate(orig, repo, node, branch'
             raise error.ProgrammingError(
                 b'largefiles is not compatible with in-memory merge'
             )
-        with lfdirstate.parentchange():
-            result = orig(repo, node, branchmerge, force, *args, **kwargs)
+        result = orig(repo, node, branchmerge, force, *args, **kwargs)
 
-            newstandins = lfutil.getstandinsstate(repo)
-            filelist = lfutil.getlfilestoupdate(oldstandins, newstandins)
+        newstandins = lfutil.getstandinsstate(repo)
+        filelist = lfutil.getlfilestoupdate(oldstandins, newstandins)
 
-            # to avoid leaving all largefiles as dirty and thus rehash them, mark
-            # all the ones that didn't change as clean
-            for lfile in oldclean.difference(filelist):
-                lfdirstate.update_file(lfile, p1_tracked=True, wc_tracked=True)
-            lfdirstate.write(repo.currenttransaction())
+        # to avoid leaving all largefiles as dirty and thus rehash them, mark
+        # all the ones that didn't change as clean
+        for lfile in oldclean.difference(filelist):
+            lfdirstate.update_file(lfile, p1_tracked=True, wc_tracked=True)
 
-            if branchmerge or force or partial:
-                filelist.extend(s.deleted + s.removed)
+        if branchmerge or force or partial:
+            filelist.extend(s.deleted + s.removed)
 
-            lfcommands.updatelfiles(
-                repo.ui, repo, filelist=filelist, normallookup=partial
-            )
+        lfcommands.updatelfiles(
+            repo.ui, repo, filelist=filelist, normallookup=partial
+        )
 
         return result
 

hgext/largefiles/reposetup.py

@@ -139,7 +139,7 b' def reposetup(ui, repo):'
             except error.LockError:
                 wlock = util.nullcontextmanager()
                 gotlock = False
-            with wlock:
+            with wlock, self.dirstate.running_status(self):
 
                 # First check if paths or patterns were specified on the
                 # command line.  If there were, and they don't match any
@@ -321,6 +321,8 b' def reposetup(ui, repo):'
 
                 if gotlock:
                     lfdirstate.write(self.currenttransaction())
+                else:
+                    lfdirstate.invalidate()
 
             self.lfstatus = True
             return scmutil.status(*result)

hgext/largefiles/storefactory.py

@@ -36,22 +36,23 b' def openstore(repo=None, remote=None, pu'
                 b'lfpullsource', repo, ui, lfpullsource
             )
         else:
-            path, _branches = urlutil.get_unique_pull_path(
-                b'lfpullsource', repo, ui, lfpullsource
+            path = urlutil.get_unique_pull_path_obj(
+                b'lfpullsource', ui, lfpullsource
             )
 
         # XXX we should not explicitly pass b'default', as this will result in
         # b'default' being returned if no `paths.default` was defined. We
         # should explicitely handle the lack of value instead.
         if repo is None:
-            path, _branches = urlutil.get_unique_pull_path(
-                b'lfs', repo, ui, b'default'
+            path = urlutil.get_unique_pull_path_obj(
+                b'lfs',
+                ui,
+                b'default',
             )
             remote = hg.peer(repo or ui, {}, path)
-        elif path == b'default-push' or path == b'default':
+        elif path.loc == b'default-push' or path.loc == b'default':
             remote = repo
         else:
-            path, _branches = urlutil.parseurl(path)
             remote = hg.peer(repo or ui, {}, path)
 
     # The path could be a scheme so use Mercurial's normal functionality

hgext/lfs/blobstore.py

@@ -168,12 +168,16 b' class local:'
             # producing the response (but the server has no way of telling us
             # that), and we really don't need to try to write the response to
             # the localstore, because it's not going to match the expected.
+            # The server also uses this method to store data uploaded by the
+            # client, so if this happens on the server side, it's possible
+            # that the client crashed or an antivirus interfered with the
+            # upload.
             if content_length is not None and int(content_length) != size:
                 msg = (
                     b"Response length (%d) does not match Content-Length "
-                    b"header (%d): likely server-side crash"
+                    b"header (%d) for %s"
                 )
-                raise LfsRemoteError(_(msg) % (size, int(content_length)))
+                raise LfsRemoteError(_(msg) % (size, int(content_length), oid))
 
             realoid = hex(sha256.digest())
             if realoid != oid:

hgext/mq.py

@@ -82,7 +82,6 b' from mercurial.pycompat import ('
 from mercurial import (
     cmdutil,
     commands,
-    dirstateguard,
     encoding,
     error,
     extensions,
@@ -791,7 +790,10 b' class queue:'
         if self.added:
             qrepo = self.qrepo()
             if qrepo:
-                qrepo[None].add(f for f in self.added if f not in qrepo[None])
+                with qrepo.wlock(), qrepo.dirstate.changing_files(qrepo):
+                    qrepo[None].add(
+                        f for f in self.added if f not in qrepo[None]
+                    )
             self.added = []
 
     def removeundo(self, repo):
@@ -1082,7 +1084,7 b' class queue:'
 
             if merge and files:
                 # Mark as removed/merged and update dirstate parent info
-                with repo.dirstate.parentchange():
+                with repo.dirstate.changing_parents(repo):
                     for f in files:
                         repo.dirstate.update_file_p1(f, p1_tracked=True)
                     p1 = repo.dirstate.p1()
@@ -1129,7 +1131,8 b' class queue:'
         if not keep:
             r = self.qrepo()
             if r:
-                r[None].forget(patches)
+                with r.wlock(), r.dirstate.changing_files(r):
+                    r[None].forget(patches)
             for p in patches:
                 try:
                     os.unlink(self.join(p))
@@ -1153,7 +1156,7 b' class queue:'
                 sortedseries.append((idx, p))
 
         sortedseries.sort(reverse=True)
-        for (i, p) in sortedseries:
+        for i, p in sortedseries:
             if i != -1:
                 del self.fullseries[i]
             else:
@@ -1177,7 +1180,6 b' class queue:'
         firstrev = repo[self.applied[0].node].rev()
         patches = []
         for i, rev in enumerate(revs):
-
             if rev < firstrev:
                 raise error.Abort(_(b'revision %d is not managed') % rev)
 
@@ -1465,7 +1467,8 b' class queue:'
                     p.close()
                     r = self.qrepo()
                     if r:
-                        r[None].add([patchfn])
+                        with r.wlock(), r.dirstate.changing_files(r):
+                            r[None].add([patchfn])
                 except:  # re-raises
                     repo.rollback()
                     raise
@@ -1830,7 +1833,7 b' class queue:'
                 if keepchanges and tobackup:
                     raise error.Abort(_(b"local changes found, qrefresh first"))
                 self.backup(repo, tobackup)
-                with repo.dirstate.parentchange():
+                with repo.dirstate.changing_parents(repo):
                     for f in a:
                         repo.wvfs.unlinkpath(f, ignoremissing=True)
                         repo.dirstate.update_file(
@@ -1988,73 +1991,67 b' class queue:'
 
             bmlist = repo[top].bookmarks()
 
-            with repo.dirstate.parentchange():
-                # XXX do we actually need the dirstateguard
-                dsguard = None
-                try:
-                    dsguard = dirstateguard.dirstateguard(repo, b'mq.refresh')
-                    if diffopts.git or diffopts.upgrade:
-                        copies = {}
-                        for dst in a:
-                            src = repo.dirstate.copied(dst)
-                            # during qfold, the source file for copies may
-                            # be removed. Treat this as a simple add.
-                            if src is not None and src in repo.dirstate:
-                                copies.setdefault(src, []).append(dst)
-                            repo.dirstate.update_file(
-                                dst, p1_tracked=False, wc_tracked=True
+            with repo.dirstate.changing_parents(repo):
+                if diffopts.git or diffopts.upgrade:
+                    copies = {}
+                    for dst in a:
+                        src = repo.dirstate.copied(dst)
+                        # during qfold, the source file for copies may
+                        # be removed. Treat this as a simple add.
+                        if src is not None and src in repo.dirstate:
+                            copies.setdefault(src, []).append(dst)
+                        repo.dirstate.update_file(
+                            dst, p1_tracked=False, wc_tracked=True
+                        )
+                    # remember the copies between patchparent and qtip
+                    for dst in aaa:
+                        src = ctx[dst].copysource()
+                        if src:
+                            copies.setdefault(src, []).extend(
+                                copies.get(dst, [])
                             )
-                        # remember the copies between patchparent and qtip
-                        for dst in aaa:
-                            src = ctx[dst].copysource()
-                            if src:
-                                copies.setdefault(src, []).extend(
-                                    copies.get(dst, [])
-                                )
-                                if dst in a:
-                                    copies[src].append(dst)
-                            # we can't copy a file created by the patch itself
-                            if dst in copies:
-                                del copies[dst]
-                        for src, dsts in copies.items():
-                            for dst in dsts:
-                                repo.dirstate.copy(src, dst)
-                    else:
-                        for dst in a:
-                            repo.dirstate.update_file(
-                                dst, p1_tracked=False, wc_tracked=True
-                            )
-                        # Drop useless copy information
-                        for f in list(repo.dirstate.copies()):
-                            repo.dirstate.copy(None, f)
-                    for f in r:
-                        repo.dirstate.update_file_p1(f, p1_tracked=True)
-                    # if the patch excludes a modified file, mark that
-                    # file with mtime=0 so status can see it.
-                    mm = []
-                    for i in range(len(m) - 1, -1, -1):
-                        if not match1(m[i]):
-                            mm.append(m[i])
-                            del m[i]
-                    for f in m:
-                        repo.dirstate.update_file_p1(f, p1_tracked=True)
-                    for f in mm:
-                        repo.dirstate.update_file_p1(f, p1_tracked=True)
-                    for f in forget:
-                        repo.dirstate.update_file_p1(f, p1_tracked=False)
-
-                    user = ph.user or ctx.user()
-
-                    oldphase = repo[top].phase()
-
-                    # assumes strip can roll itself back if interrupted
-                    repo.setparents(*cparents)
-                    self.applied.pop()
-                    self.applieddirty = True
-                    strip(self.ui, repo, [top], update=False, backup=False)
-                    dsguard.close()
-                finally:
-                    release(dsguard)
+                            if dst in a:
+                                copies[src].append(dst)
+                        # we can't copy a file created by the patch itself
+                        if dst in copies:
+                            del copies[dst]
+                    for src, dsts in copies.items():
+                        for dst in dsts:
+                            repo.dirstate.copy(src, dst)
+                else:
+                    for dst in a:
+                        repo.dirstate.update_file(
+                            dst, p1_tracked=False, wc_tracked=True
+                        )
+                    # Drop useless copy information
+                    for f in list(repo.dirstate.copies()):
+                        repo.dirstate.copy(None, f)
+                for f in r:
+                    repo.dirstate.update_file_p1(f, p1_tracked=True)
+                # if the patch excludes a modified file, mark that
+                # file with mtime=0 so status can see it.
+                mm = []
+                for i in range(len(m) - 1, -1, -1):
+                    if not match1(m[i]):
+                        mm.append(m[i])
+                        del m[i]
+                for f in m:
+                    repo.dirstate.update_file_p1(f, p1_tracked=True)
+                for f in mm:
+                    repo.dirstate.update_file_p1(f, p1_tracked=True)
+                for f in forget:
+                    repo.dirstate.update_file_p1(f, p1_tracked=False)
+
+                user = ph.user or ctx.user()
+
+                oldphase = repo[top].phase()
+
+                # assumes strip can roll itself back if interrupted
+                repo.setparents(*cparents)
+                repo.dirstate.write(repo.currenttransaction())
+                self.applied.pop()
+                self.applieddirty = True
+                strip(self.ui, repo, [top], update=False, backup=False)
 
             try:
                 # might be nice to attempt to roll back strip after this
@@ -2124,8 +2121,9 b' class queue:'
                 finally:
                     lockmod.release(tr, lock)
             except:  # re-raises
-                ctx = repo[cparents[0]]
-                repo.dirstate.rebuild(ctx.node(), ctx.manifest())
+                with repo.dirstate.changing_parents(repo):
+                    ctx = repo[cparents[0]]
+                    repo.dirstate.rebuild(ctx.node(), ctx.manifest())
                 self.savedirty()
                 self.ui.warn(
                     _(
@@ -2760,18 +2758,19 b' def qinit(ui, repo, create):'
     r = q.init(repo, create)
     q.savedirty()
     if r:
-        if not os.path.exists(r.wjoin(b'.hgignore')):
-            fp = r.wvfs(b'.hgignore', b'w')
-            fp.write(b'^\\.hg\n')
-            fp.write(b'^\\.mq\n')
-            fp.write(b'syntax: glob\n')
-            fp.write(b'status\n')
-            fp.write(b'guards\n')
-            fp.close()
-        if not os.path.exists(r.wjoin(b'series')):
-            r.wvfs(b'series', b'w').close()
-        r[None].add([b'.hgignore', b'series'])
-        commands.add(ui, r)
+        with r.wlock(), r.dirstate.changing_files(r):
+            if not os.path.exists(r.wjoin(b'.hgignore')):
+                fp = r.wvfs(b'.hgignore', b'w')
+                fp.write(b'^\\.hg\n')
+                fp.write(b'^\\.mq\n')
+                fp.write(b'syntax: glob\n')
+                fp.write(b'status\n')
+                fp.write(b'guards\n')
+                fp.close()
+            if not os.path.exists(r.wjoin(b'series')):
+                r.wvfs(b'series', b'w').close()
+            r[None].add([b'.hgignore', b'series'])
+            commands.add(ui, r)
     return 0
 
 
@@ -2854,16 +2853,17 b' def clone(ui, source, dest=None, **opts)'
     # main repo (destination and sources)
     if dest is None:
         dest = hg.defaultdest(source)
-    __, source_path, __ = urlutil.get_clone_path(ui, source)
+    source_path = urlutil.get_clone_path_obj(ui, source)
     sr = hg.peer(ui, opts, source_path)
 
     # patches repo (source only)
     if opts.get(b'patches'):
-        __, patchespath, __ = urlutil.get_clone_path(ui, opts.get(b'patches'))
+        patches_path = urlutil.get_clone_path_obj(ui, opts.get(b'patches'))
     else:
-        patchespath = patchdir(sr)
+        # XXX path: we should turn this into a path object
+        patches_path = patchdir(sr)
     try:
-        hg.peer(ui, opts, patchespath)
+        hg.peer(ui, opts, patches_path)
     except error.RepoError:
         raise error.Abort(
             _(b'versioned patch repository not found (see init --mq)')
@@ -3223,45 +3223,46 b' def fold(ui, repo, *files, **opts):'
         raise error.Abort(_(b'qfold requires at least one patch name'))
     if not q.checktoppatch(repo)[0]:
         raise error.Abort(_(b'no patches applied'))
-    q.checklocalchanges(repo)
-
-    message = cmdutil.logmessage(ui, opts)
-
-    parent = q.lookup(b'qtip')
-    patches = []
-    messages = []
-    for f in files:
-        p = q.lookup(f)
-        if p in patches or p == parent:
-            ui.warn(_(b'skipping already folded patch %s\n') % p)
-        if q.isapplied(p):
-            raise error.Abort(
-                _(b'qfold cannot fold already applied patch %s') % p
-            )
-        patches.append(p)
-
-    for p in patches:
+
+    with repo.wlock():
+        q.checklocalchanges(repo)
+
+        message = cmdutil.logmessage(ui, opts)
+
+        parent = q.lookup(b'qtip')
+        patches = []
+        messages = []
+        for f in files:
+            p = q.lookup(f)
+            if p in patches or p == parent:
+                ui.warn(_(b'skipping already folded patch %s\n') % p)
+            if q.isapplied(p):
+                raise error.Abort(
+                    _(b'qfold cannot fold already applied patch %s') % p
+                )
+            patches.append(p)
+
+        for p in patches:
+            if not message:
+                ph = patchheader(q.join(p), q.plainmode)
+                if ph.message:
+                    messages.append(ph.message)
+            pf = q.join(p)
+            (patchsuccess, files, fuzz) = q.patch(repo, pf)
+            if not patchsuccess:
+                raise error.Abort(_(b'error folding patch %s') % p)
+
         if not message:
-            ph = patchheader(q.join(p), q.plainmode)
-            if ph.message:
-                messages.append(ph.message)
-        pf = q.join(p)
-        (patchsuccess, files, fuzz) = q.patch(repo, pf)
-        if not patchsuccess:
-            raise error.Abort(_(b'error folding patch %s') % p)
-
-    if not message:
-        ph = patchheader(q.join(parent), q.plainmode)
-        message = ph.message
-        for msg in messages:
-            if msg:
-                if message:
-                    message.append(b'* * *')
-                message.extend(msg)
-        message = b'\n'.join(message)
-
-    diffopts = q.patchopts(q.diffopts(), *patches)
-    with repo.wlock():
+            ph = patchheader(q.join(parent), q.plainmode)
+            message = ph.message
+            for msg in messages:
+                if msg:
+                    if message:
+                        message.append(b'* * *')
+                    message.extend(msg)
+            message = b'\n'.join(message)
+
+        diffopts = q.patchopts(q.diffopts(), *patches)
         q.refresh(
             repo,
             msg=message,
@@ -3627,8 +3628,8 b' def rename(ui, repo, patch, name=None, *'
     util.rename(q.join(patch), absdest)
     r = q.qrepo()
     if r and patch in r.dirstate:
-        wctx = r[None]
-        with r.wlock():
+        with r.wlock(), r.dirstate.changing_files(r):
+            wctx = r[None]
             if r.dirstate.get_entry(patch).added:
                 r.dirstate.set_untracked(patch)
                 r.dirstate.set_tracked(name)

hgext/narrow/narrowcommands.py

@@ -320,7 +320,7 b' def _narrow('
                 repo.store.markremoved(f)
 
             ui.status(_(b'deleting unwanted files from working copy\n'))
-            with repo.dirstate.parentchange():
+            with repo.dirstate.changing_parents(repo):
                 narrowspec.updateworkingcopy(repo, assumeclean=True)
                 narrowspec.copytoworkingcopy(repo)
 
@@ -380,7 +380,7 b' def _widen('
         if ellipsesremote:
             ds = repo.dirstate
             p1, p2 = ds.p1(), ds.p2()
-            with ds.parentchange():
+            with ds.changing_parents(repo):
                 ds.setparents(repo.nullid, repo.nullid)
         if isoldellipses:
             with wrappedextraprepare:
@@ -416,13 +416,15 b' def _widen('
                     repo, trmanager.transaction, source=b'widen'
                 )
                 # TODO: we should catch error.Abort here
-                bundle2.processbundle(repo, bundle, op=op)
+                bundle2.processbundle(repo, bundle, op=op, remote=remote)
 
         if ellipsesremote:
-            with ds.parentchange():
+            with ds.changing_parents(repo):
                 ds.setparents(p1, p2)
 
-        with repo.transaction(b'widening'), repo.dirstate.parentchange():
+        with repo.transaction(b'widening'), repo.dirstate.changing_parents(
+            repo
+        ):
             repo.setnewnarrowpats()
             narrowspec.updateworkingcopy(repo)
             narrowspec.copytoworkingcopy(repo)
@@ -591,7 +593,7 b' def trackedcmd(ui, repo, remotepath=None'
     if update_working_copy:
         with repo.wlock(), repo.lock(), repo.transaction(
             b'narrow-wc'
-        ), repo.dirstate.parentchange():
+        ), repo.dirstate.changing_parents(repo):
             narrowspec.updateworkingcopy(repo)
             narrowspec.copytoworkingcopy(repo)
         return 0
@@ -606,10 +608,9 b' def trackedcmd(ui, repo, remotepath=None'
         # Find the revisions we have in common with the remote. These will
         # be used for finding local-only changes for narrowing. They will
         # also define the set of revisions to update for widening.
-        r = urlutil.get_unique_pull_path(b'tracked', repo, ui, remotepath)
-        url, branches = r
-        ui.status(_(b'comparing with %s\n') % urlutil.hidepassword(url))
-        remote = hg.peer(repo, opts, url)
+        path = urlutil.get_unique_pull_path_obj(b'tracked', ui, remotepath)
+        ui.status(_(b'comparing with %s\n') % urlutil.hidepassword(path.loc))
+        remote = hg.peer(repo, opts, path)
 
         try:
             # check narrow support before doing anything if widening needs to be

hgext/narrow/narrowrepo.py

@@ -19,8 +19,8 b' def wraprepo(repo):'
             dirstate = super(narrowrepository, self)._makedirstate()
             return narrowdirstate.wrapdirstate(self, dirstate)
 
-        def peer(self):
-            peer = super(narrowrepository, self).peer()
+        def peer(self, path=None):
+            peer = super(narrowrepository, self).peer(path=path)
             peer._caps.add(wireprototypes.NARROWCAP)
             peer._caps.add(wireprototypes.ELLIPSESCAP)
             return peer

hgext/notify.py

@@ -450,7 +450,7 b' class notifier:'
         try:
             msg = mail.parsebytes(data)
         except emailerrors.MessageParseError as inst:
-            raise error.Abort(inst)
+            raise error.Abort(stringutil.forcebytestr(inst))
 
         # store sender and subject
         sender = msg['From']

hgext/phabricator.py

@@ -286,9 +286,12 b' def vcrcommand(name, flags, spec, helpca'
                 import hgdemandimport
 
                 with hgdemandimport.deactivated():
+                    # pytype: disable=import-error
                     import vcr as vcrmod
                     import vcr.stubs as stubs
 
+                    # pytype: enable=import-error
+
                     vcr = vcrmod.VCR(
                         serializer='json',
                         before_record_request=sanitiserequest,
@@ -350,11 +353,14 b' def urlencodenested(params):'
     """
     flatparams = util.sortdict()
 
-    def process(prefix, obj):
+    def process(prefix: bytes, obj):
         if isinstance(obj, bool):
             obj = {True: b'true', False: b'false'}[obj]  # Python -> PHP form
         lister = lambda l: [(b'%d' % k, v) for k, v in enumerate(l)]
+        # .items() will only be called for a dict type
+        # pytype: disable=attribute-error
         items = {list: lister, dict: lambda x: x.items()}.get(type(obj))
+        # pytype: enable=attribute-error
         if items is None:
             flatparams[prefix] = obj
         else:

hgext/rebase.py

@@ -30,7 +30,6 b' from mercurial import ('
     commands,
     copies,
     destutil,
-    dirstateguard,
     error,
     extensions,
     logcmdutil,
@@ -1271,15 +1270,9 b' def _origrebase(ui, repo, action, opts, '
         # one transaction here. Otherwise, transactions are obtained when
         # committing each node, which is slower but allows partial success.
         with util.acceptintervention(tr):
-            # Same logic for the dirstate guard, except we don't create one when
-            # rebasing in-memory (it's not needed).
-            dsguard = None
-            if singletr and not rbsrt.inmemory:
-                dsguard = dirstateguard.dirstateguard(repo, b'rebase')
-            with util.acceptintervention(dsguard):
-                rbsrt._performrebase(tr)
-                if not rbsrt.dryrun:
-                    rbsrt._finishrebase()
+            rbsrt._performrebase(tr)
+            if not rbsrt.dryrun:
+                rbsrt._finishrebase()
 
 
 def _definedestmap(ui, repo, inmemory, destf, srcf, basef, revf, destspace):
@@ -1500,10 +1493,10 b' def commitmemorynode(repo, wctx, editor,'
 def commitnode(repo, editor, extra, user, date, commitmsg):
     """Commit the wd changes with parents p1 and p2.
     Return node of committed revision."""
-    dsguard = util.nullcontextmanager()
+    tr = util.nullcontextmanager
     if not repo.ui.configbool(b'rebase', b'singletransaction'):
-        dsguard = dirstateguard.dirstateguard(repo, b'rebase')
-    with dsguard:
+        tr = lambda: repo.transaction(b'rebase')
+    with tr():
         # Commit might fail if unresolved files exist
         newnode = repo.commit(
             text=commitmsg, user=user, date=date, extra=extra, editor=editor
@@ -1520,12 +1513,14 b' def rebasenode(repo, rev, p1, p2, base, '
     p1ctx = repo[p1]
     if wctx.isinmemory():
         wctx.setbase(p1ctx)
+        scope = util.nullcontextmanager
     else:
         if repo[b'.'].rev() != p1:
             repo.ui.debug(b" update to %d:%s\n" % (p1, p1ctx))
             mergemod.clean_update(p1ctx)
         else:
             repo.ui.debug(b" already in destination\n")
+        scope = lambda: repo.dirstate.changing_parents(repo)
         # This is, alas, necessary to invalidate workingctx's manifest cache,
         # as well as other data we litter on it in other places.
         wctx = repo[None]
@@ -1535,26 +1530,27 b' def rebasenode(repo, rev, p1, p2, base, '
     if base is not None:
         repo.ui.debug(b"   detach base %d:%s\n" % (base, repo[base]))
 
-    # See explanation in merge.graft()
-    mergeancestor = repo.changelog.isancestor(p1ctx.node(), ctx.node())
-    stats = mergemod._update(
-        repo,
-        rev,
-        branchmerge=True,
-        force=True,
-        ancestor=base,
-        mergeancestor=mergeancestor,
-        labels=[b'dest', b'source', b'parent of source'],
-        wc=wctx,
-    )
-    wctx.setparents(p1ctx.node(), repo[p2].node())
-    if collapse:
-        copies.graftcopies(wctx, ctx, p1ctx)
-    else:
-        # If we're not using --collapse, we need to
-        # duplicate copies between the revision we're
-        # rebasing and its first parent.
-        copies.graftcopies(wctx, ctx, ctx.p1())
+    with scope():
+        # See explanation in merge.graft()
+        mergeancestor = repo.changelog.isancestor(p1ctx.node(), ctx.node())
+        stats = mergemod._update(
+            repo,
+            rev,
+            branchmerge=True,
+            force=True,
+            ancestor=base,
+            mergeancestor=mergeancestor,
+            labels=[b'dest', b'source', b'parent of source'],
+            wc=wctx,
+        )
+        wctx.setparents(p1ctx.node(), repo[p2].node())
+        if collapse:
+            copies.graftcopies(wctx, ctx, p1ctx)
+        else:
+            # If we're not using --collapse, we need to
+            # duplicate copies between the revision we're
+            # rebasing and its first parent.
+            copies.graftcopies(wctx, ctx, ctx.p1())
 
     if stats.unresolvedcount > 0:
         if wctx.isinmemory():

hgext/releasenotes.py

@@ -39,7 +39,7 b' command = registrar.command(cmdtable)'
 try:
     # Silence a warning about python-Levenshtein.
     #
-    # We don't need the the performance that much and it get anoying in tests.
+    # We don't need the performance that much and it gets annoying in tests.
     import warnings
 
     with warnings.catch_warnings():
@@ -50,7 +50,7 b' try:'
             module="fuzzywuzzy.fuzz",
         )
 
-        import fuzzywuzzy.fuzz as fuzz
+        import fuzzywuzzy.fuzz as fuzz  # pytype: disable=import-error
 
         fuzz.token_set_ratio
 except ImportError:

hgext/relink.py

@@ -67,8 +67,8 b' def relink(ui, repo, origin=None, **opts'
 
     if origin is None and b'default-relink' in ui.paths:
         origin = b'default-relink'
-    path, __ = urlutil.get_unique_pull_path(b'relink', repo, ui, origin)
-    src = hg.repository(repo.baseui, path)
+    path = urlutil.get_unique_pull_path_obj(b'relink', ui, origin)
+    src = hg.repository(repo.baseui, path.loc)
     ui.status(_(b'relinking %s to %s\n') % (src.store.path, repo.store.path))
     if repo.root == src.root:
         ui.status(_(b'there is nothing to relink\n'))

hgext/remotefilelog/remotefilelog.py

@@ -299,6 +299,7 b' class remotefilelog:'
         deltaprevious=False,
         deltamode=None,
         sidedata_helpers=None,
+        debug_info=None,
     ):
         # we don't use any of these parameters here
         del nodesorder, revisiondata, assumehaveparentrevisions, deltaprevious

hgext/remotefilelog/shallowutil.py

@@ -247,7 +247,7 b' def parsesizeflags(raw):'
         index = raw.index(b'\0')
     except ValueError:
         raise BadRemotefilelogHeader(
-            "unexpected remotefilelog header: illegal format"
+            b"unexpected remotefilelog header: illegal format"
         )
     header = raw[:index]
     if header.startswith(b'v'):
@@ -267,7 +267,7 b' def parsesizeflags(raw):'
         size = int(header)
     if size is None:
         raise BadRemotefilelogHeader(
-            "unexpected remotefilelog header: no size found"
+            b"unexpected remotefilelog header: no size found"
         )
     return index + 1, size, flags
 

hgext/schemes.py

@@ -80,9 +80,25 b' class ShortRepository:'
     def __repr__(self):
         return b'<ShortRepository: %s>' % self.scheme
 
+    def make_peer(self, ui, path, *args, **kwargs):
+        new_url = self.resolve(path.rawloc)
+        path = path.copy(new_raw_location=new_url)
+        cls = hg.peer_schemes.get(path.url.scheme)
+        if cls is not None:
+            return cls.make_peer(ui, path, *args, **kwargs)
+        return None
+
     def instance(self, ui, url, create, intents=None, createopts=None):
         url = self.resolve(url)
-        return hg._peerlookup(url).instance(
+        u = urlutil.url(url)
+        scheme = u.scheme or b'file'
+        if scheme in hg.peer_schemes:
+            cls = hg.peer_schemes[scheme]
+        elif scheme in hg.repo_schemes:
+            cls = hg.repo_schemes[scheme]
+        else:
+            cls = hg.LocalFactory
+        return cls.instance(
             ui, url, create, intents=intents, createopts=createopts
         )
 
@@ -119,24 +135,29 b' schemes = {'
 }
 
 
+def _check_drive_letter(scheme: bytes) -> None:
+    """check if a scheme conflict with a Windows drive letter"""
+    if (
+        pycompat.iswindows
+        and len(scheme) == 1
+        and scheme.isalpha()
+        and os.path.exists(b'%s:\\' % scheme)
+    ):
+        msg = _(b'custom scheme %s:// conflicts with drive letter %s:\\\n')
+        msg %= (scheme, scheme.upper())
+        raise error.Abort(msg)
+
+
 def extsetup(ui):
     schemes.update(dict(ui.configitems(b'schemes')))
     t = templater.engine(templater.parse)
     for scheme, url in schemes.items():
-        if (
-            pycompat.iswindows
-            and len(scheme) == 1
-            and scheme.isalpha()
-            and os.path.exists(b'%s:\\' % scheme)
-        ):
-            raise error.Abort(
-                _(
-                    b'custom scheme %s:// conflicts with drive '
-                    b'letter %s:\\\n'
-                )
-                % (scheme, scheme.upper())
-            )
-        hg.schemes[scheme] = ShortRepository(url, scheme, t)
+        _check_drive_letter(scheme)
+        url_scheme = urlutil.url(url).scheme
+        if url_scheme in hg.peer_schemes:
+            hg.peer_schemes[scheme] = ShortRepository(url, scheme, t)
+        else:
+            hg.repo_schemes[scheme] = ShortRepository(url, scheme, t)
 
     extensions.wrapfunction(urlutil, b'hasdriveletter', hasdriveletter)
 
@@ -144,7 +165,11 b' def extsetup(ui):'
 @command(b'debugexpandscheme', norepo=True)
 def expandscheme(ui, url, **opts):
     """given a repo path, provide the scheme-expanded path"""
-    repo = hg._peerlookup(url)
-    if isinstance(repo, ShortRepository):
-        url = repo.resolve(url)
+    scheme = urlutil.url(url).scheme
+    if scheme in hg.peer_schemes:
+        cls = hg.peer_schemes[scheme]
+    else:
+        cls = hg.repo_schemes.get(scheme)
+    if cls is not None and isinstance(cls, ShortRepository):
+        url = cls.resolve(url)
     ui.write(url + b'\n')

hgext/split.py

@@ -134,7 +134,7 b' def dosplit(ui, repo, tr, ctx, opts):'
     # Set working parent to ctx.p1(), and keep working copy as ctx's content
     if ctx.node() != repo.dirstate.p1():
         hg.clean(repo, ctx.node(), show_stats=False)
-    with repo.dirstate.parentchange():
+    with repo.dirstate.changing_parents(repo):
         scmutil.movedirstate(repo, ctx.p1())
 
     # Any modified, added, removed, deleted result means split is incomplete

hgext/sqlitestore.py

@@ -80,7 +80,7 b' from mercurial.utils import ('
 )
 
 try:
-    from mercurial import zstd
+    from mercurial import zstd  # pytype: disable=import-error
 
     zstd.__version__
 except ImportError:
@@ -608,6 +608,7 b' class sqlitefilestore:'
         assumehaveparentrevisions=False,
         deltamode=repository.CG_DELTAMODE_STD,
         sidedata_helpers=None,
+        debug_info=None,
     ):
         if nodesorder not in (b'nodes', b'storage', b'linear', None):
             raise error.ProgrammingError(

hgext/transplant.py

@@ -817,8 +817,8 b' def _dotransplant(ui, repo, *revs, **opt'
 
     sourcerepo = opts.get(b'source')
     if sourcerepo:
-        u = urlutil.get_unique_pull_path(b'transplant', repo, ui, sourcerepo)[0]
-        peer = hg.peer(repo, opts, u)
+        path = urlutil.get_unique_pull_path_obj(b'transplant', ui, sourcerepo)
+        peer = hg.peer(repo, opts, path)
         heads = pycompat.maplist(peer.lookup, opts.get(b'branch', ()))
         target = set(heads)
         for r in revs:

hgext/uncommit.py

@@ -236,7 +236,7 b' def uncommit(ui, repo, *pats, **opts):'
                 # Fully removed the old commit
                 mapping[old.node()] = ()
 
-            with repo.dirstate.parentchange():
+            with repo.dirstate.changing_parents(repo):
                 scmutil.movedirstate(repo, repo[newid], match)
 
             scmutil.cleanupnodes(repo, mapping, b'uncommit', fixphase=True)
@@ -317,7 +317,7 b' def unamend(ui, repo, **opts):'
         newpredctx = repo[newprednode]
         dirstate = repo.dirstate
 
-        with dirstate.parentchange():
+        with dirstate.changing_parents(repo):
             scmutil.movedirstate(repo, newpredctx)
 
         mapping = {curctx.node(): (newprednode,)}

hgext/win32text.py

@@ -216,17 +216,23 b' def reposetup(ui, repo):'
 def wrap_revert(orig, repo, ctx, names, uipathfn, actions, *args, **kwargs):
     # reset dirstate cache for file we touch
     ds = repo.dirstate
-    with ds.parentchange():
-        for filename in actions[b'revert'][0]:
-            entry = ds.get_entry(filename)
-            if entry is not None:
-                if entry.p1_tracked:
-                    ds.update_file(
-                        filename,
-                        entry.tracked,
-                        p1_tracked=True,
-                        p2_info=entry.p2_info,
-                    )
+    for filename in actions[b'revert'][0]:
+        entry = ds.get_entry(filename)
+        if entry is not None:
+            if entry.p1_tracked:
+                # If we revert the file, it is possibly dirty. However,
+                # this extension meddle with the file content and therefore
+                # its size. As a result, we cannot simply call
+                # `dirstate.set_possibly_dirty` as it will not affet the
+                # expected size of the file.
+                #
+                # At least, now, the quirk is properly documented.
+                ds.hacky_extension_update_file(
+                    filename,
+                    entry.tracked,
+                    p1_tracked=entry.p1_tracked,
+                    p2_info=entry.p2_info,
+                )
     return orig(repo, ctx, names, uipathfn, actions, *args, **kwargs)
 
 

mercurial/archival.py

@@ -154,9 +154,14 b' class tarit:'
                 )
                 self.fileobj = gzfileobj
                 return (
+                    # taropen() wants Literal['a', 'r', 'w', 'x'] for the mode,
+                    # but Literal[] is only available in 3.8+ without the
+                    # typing_extensions backport.
+                    # pytype: disable=wrong-arg-types
                     tarfile.TarFile.taropen(  # pytype: disable=attribute-error
                         name, pycompat.sysstr(mode), gzfileobj
                     )
+                    # pytype: enable=wrong-arg-types
                 )
             else:
                 try:

mercurial/bundle2.py

@@ -315,8 +315,17 b' class bundleoperation:'
     * a way to construct a bundle response when applicable.
     """
 
-    def __init__(self, repo, transactiongetter, captureoutput=True, source=b''):
+    def __init__(
+        self,
+        repo,
+        transactiongetter,
+        captureoutput=True,
+        source=b'',
+        remote=None,
+    ):
         self.repo = repo
+        # the peer object who produced this bundle if available
+        self.remote = remote
         self.ui = repo.ui
         self.records = unbundlerecords()
         self.reply = None
@@ -363,7 +372,7 b' def _notransaction():'
     raise TransactionUnavailable()
 
 
-def applybundle(repo, unbundler, tr, source, url=None, **kwargs):
+def applybundle(repo, unbundler, tr, source, url=None, remote=None, **kwargs):
     # transform me into unbundler.apply() as soon as the freeze is lifted
     if isinstance(unbundler, unbundle20):
         tr.hookargs[b'bundle2'] = b'1'
@@ -371,10 +380,12 b' def applybundle(repo, unbundler, tr, sou'
             tr.hookargs[b'source'] = source
         if url is not None and b'url' not in tr.hookargs:
             tr.hookargs[b'url'] = url
-        return processbundle(repo, unbundler, lambda: tr, source=source)
+        return processbundle(
+            repo, unbundler, lambda: tr, source=source, remote=remote
+        )
     else:
         # the transactiongetter won't be used, but we might as well set it
-        op = bundleoperation(repo, lambda: tr, source=source)
+        op = bundleoperation(repo, lambda: tr, source=source, remote=remote)
         _processchangegroup(op, unbundler, tr, source, url, **kwargs)
         return op
 
@@ -450,7 +461,14 b' class partiterator:'
         )
 
 
-def processbundle(repo, unbundler, transactiongetter=None, op=None, source=b''):
+def processbundle(
+    repo,
+    unbundler,
+    transactiongetter=None,
+    op=None,
+    source=b'',
+    remote=None,
+):
     """This function process a bundle, apply effect to/from a repo
 
     It iterates over each part then searches for and uses the proper handling
@@ -466,7 +484,12 b' def processbundle(repo, unbundler, trans'
     if op is None:
         if transactiongetter is None:
             transactiongetter = _notransaction
-        op = bundleoperation(repo, transactiongetter, source=source)
+        op = bundleoperation(
+            repo,
+            transactiongetter,
+            source=source,
+            remote=remote,
+        )
     # todo:
     # - replace this is a init function soon.
     # - exception catching
@@ -494,6 +517,10 b' def processparts(repo, op, unbundler):'
 
 
 def _processchangegroup(op, cg, tr, source, url, **kwargs):
+    if op.remote is not None and op.remote.path is not None:
+        remote_path = op.remote.path
+        kwargs = kwargs.copy()
+        kwargs['delta_base_reuse_policy'] = remote_path.delta_reuse_policy
     ret = cg.apply(op.repo, tr, source, url, **kwargs)
     op.records.add(
         b'changegroup',
@@ -1938,7 +1965,12 b' def writebundle('
             raise error.Abort(
                 _(b'old bundle types only supports v1 changegroups')
             )
+
+        # HG20 is the case without 2 values to unpack, but is handled above.
+        # pytype: disable=bad-unpacking
         header, comp = bundletypes[bundletype]
+        # pytype: enable=bad-unpacking
+
         if comp not in util.compengines.supportedbundletypes:
             raise error.Abort(_(b'unknown stream compression type: %s') % comp)
         compengine = util.compengines.forbundletype(comp)

mercurial/bundlecaches.py

@@ -5,6 +5,10 b''
 
 import collections
 
+from typing import (
+    cast,
+)
+
 from .i18n import _
 
 from .thirdparty import attr
@@ -247,7 +251,7 b' def parsebundlespec(repo, spec, strict=T'
     # required to apply it. If we see this metadata, compare against what the
     # repo supports and error if the bundle isn't compatible.
     if version == b'packed1' and b'requirements' in params:
-        requirements = set(params[b'requirements'].split(b','))
+        requirements = set(cast(bytes, params[b'requirements']).split(b','))
         missingreqs = requirements - requirementsmod.STREAM_FIXED_REQUIREMENTS
         if missingreqs:
             raise error.UnsupportedBundleSpecification(

mercurial/bundlerepo.py

@@ -88,7 +88,7 b' class bundlerevlog(revlog.revlog):'
                     )
 
             if not self.index.has_node(deltabase):
-                raise LookupError(
+                raise error.LookupError(
                     deltabase, self.display_id, _(b'unknown delta base')
                 )
 
@@ -458,8 +458,8 b' class bundlerepository:'
     def cancopy(self):
         return False
 
-    def peer(self):
-        return bundlepeer(self)
+    def peer(self, path=None):
+        return bundlepeer(self, path=path)
 
     def getcwd(self):
         return encoding.getcwd()  # always outside the repo

mercurial/cext/bdiff.pyi

@@ -5,7 +5,7 b' from typing import ('
 
 version: int
 
-def bdiff(a: bytes, b: bytes): bytes
+def bdiff(a: bytes, b: bytes) -> bytes: ...
 def blocks(a: bytes, b: bytes) -> List[Tuple[int, int, int, int]]: ...
 def fixws(s: bytes, allws: bool) -> bytes: ...
 def splitnewlines(text: bytes) -> List[bytes]: ...

mercurial/cext/osutil.pyi

@@ -2,6 +2,7 b' from typing import ('
     AnyStr,
     IO,
     List,
+    Optional,
     Sequence,
 )
 
@@ -15,7 +16,7 b' class stat:'
     st_mtime: int
     st_ctime: int
 
-def listdir(path: bytes, st: bool, skip: bool) -> List[stat]: ...
+def listdir(path: bytes, st: bool, skip: Optional[bool]) -> List[stat]: ...
 def posixfile(name: AnyStr, mode: bytes, buffering: int) -> IO: ...
 def statfiles(names: Sequence[bytes]) -> List[stat]: ...
 def setprocname(name: bytes) -> None: ...

mercurial/cext/parsers.c

@@ -177,7 +177,7 b' static inline bool dirstate_item_c_remov'
 	        (dirstate_flag_p1_tracked | dirstate_flag_p2_info));
 }
 
-static inline bool dirstate_item_c_merged(dirstateItemObject *self)
+static inline bool dirstate_item_c_modified(dirstateItemObject *self)
 {
 	return ((self->flags & dirstate_flag_wc_tracked) &&
 	        (self->flags & dirstate_flag_p1_tracked) &&
@@ -195,7 +195,7 b' static inline char dirstate_item_c_v1_st'
 {
 	if (dirstate_item_c_removed(self)) {
 		return 'r';
-	} else if (dirstate_item_c_merged(self)) {
+	} else if (dirstate_item_c_modified(self)) {
 		return 'm';
 	} else if (dirstate_item_c_added(self)) {
 		return 'a';
@@ -642,9 +642,9 b' static PyObject *dirstate_item_get_p2_in'
 	}
 };
 
-static PyObject *dirstate_item_get_merged(dirstateItemObject *self)
+static PyObject *dirstate_item_get_modified(dirstateItemObject *self)
 {
-	if (dirstate_item_c_merged(self)) {
+	if (dirstate_item_c_modified(self)) {
 		Py_RETURN_TRUE;
 	} else {
 		Py_RETURN_FALSE;
@@ -709,7 +709,7 b' static PyGetSetDef dirstate_item_getset['
      NULL},
     {"added", (getter)dirstate_item_get_added, NULL, "added", NULL},
     {"p2_info", (getter)dirstate_item_get_p2_info, NULL, "p2_info", NULL},
-    {"merged", (getter)dirstate_item_get_merged, NULL, "merged", NULL},
+    {"modified", (getter)dirstate_item_get_modified, NULL, "modified", NULL},
     {"from_p2", (getter)dirstate_item_get_from_p2, NULL, "from_p2", NULL},
     {"maybe_clean", (getter)dirstate_item_get_maybe_clean, NULL, "maybe_clean",
      NULL},
@@ -1187,7 +1187,7 b' void dirs_module_init(PyObject *mod);'
 void manifest_module_init(PyObject *mod);
 void revlog_module_init(PyObject *mod);
 
-static const int version = 20;
+static const int version = 21;
 
 static void module_init(PyObject *mod)
 {

mercurial/cext/parsers.pyi

@@ -76,3 +76,7 b' class nodetree:'
 
     def insert(self, rev: int) -> None: ...
     def shortest(self, node: bytes) -> int: ...
+
+# The IndexObject type here is defined in C, and there's no type for a buffer
+# return, as of py3.11.  https://github.com/python/typing/issues/593
+def parse_index2(data: object, inline: object, format: int = ...) -> Tuple[object, Optional[Tuple[int, object]]]: ...

mercurial/cext/revlog.c

@@ -1446,16 +1446,25 b' static PyObject *index_issnapshot(indexO'
 static PyObject *index_findsnapshots(indexObject *self, PyObject *args)
 {
 	Py_ssize_t start_rev;
+	Py_ssize_t end_rev;
 	PyObject *cache;
 	Py_ssize_t base;
 	Py_ssize_t rev;
 	PyObject *key = NULL;
 	PyObject *value = NULL;
 	const Py_ssize_t length = index_length(self);
-	if (!PyArg_ParseTuple(args, "O!n", &PyDict_Type, &cache, &start_rev)) {
+	if (!PyArg_ParseTuple(args, "O!nn", &PyDict_Type, &cache, &start_rev,
+	                      &end_rev)) {
 		return NULL;
 	}
-	for (rev = start_rev; rev < length; rev++) {
+	end_rev += 1;
+	if (end_rev > length) {
+		end_rev = length;
+	}
+	if (start_rev < 0) {
+		start_rev = 0;
+	}
+	for (rev = start_rev; rev < end_rev; rev++) {
 		int issnap;
 		PyObject *allvalues = NULL;
 		issnap = index_issnapshotrev(self, rev);
@@ -1480,7 +1489,7 b' static PyObject *index_findsnapshots(ind'
 		}
 		if (allvalues == NULL) {
 			int r;
-			allvalues = PyList_New(0);
+			allvalues = PySet_New(0);
 			if (!allvalues) {
 				goto bail;
 			}
@@ -1491,7 +1500,7 b' static PyObject *index_findsnapshots(ind'
 			}
 		}
 		value = PyLong_FromSsize_t(rev);
-		if (PyList_Append(allvalues, value)) {
+		if (PySet_Add(allvalues, value)) {
 			goto bail;
 		}
 		Py_CLEAR(key);

mercurial/cffi/bdiff.py

@@ -8,6 +8,11 b''
 
 import struct
 
+from typing import (
+    List,
+    Tuple,
+)
+
 from ..pure.bdiff import *
 from . import _bdiff  # pytype: disable=import-error
 
@@ -15,7 +20,7 b' ffi = _bdiff.ffi'
 lib = _bdiff.lib
 
 
-def blocks(sa, sb):
+def blocks(sa: bytes, sb: bytes) -> List[Tuple[int, int, int, int]]:
     a = ffi.new(b"struct bdiff_line**")
     b = ffi.new(b"struct bdiff_line**")
     ac = ffi.new(b"char[]", str(sa))
@@ -29,7 +34,7 b' def blocks(sa, sb):'
         count = lib.bdiff_diff(a[0], an, b[0], bn, l)
         if count < 0:
             raise MemoryError
-        rl = [None] * count
+        rl = [(0, 0, 0, 0)] * count
         h = l.next
         i = 0
         while h:
@@ -43,7 +48,7 b' def blocks(sa, sb):'
     return rl
 
 
-def bdiff(sa, sb):
+def bdiff(sa: bytes, sb: bytes) -> bytes:
     a = ffi.new(b"struct bdiff_line**")
     b = ffi.new(b"struct bdiff_line**")
     ac = ffi.new(b"char[]", str(sa))

mercurial/cffi/mpatch.py

@@ -6,6 +6,8 b''
 # GNU General Public License version 2 or any later version.
 
 
+from typing import List
+
 from ..pure.mpatch import *
 from ..pure.mpatch import mpatchError  # silence pyflakes
 from . import _mpatch  # pytype: disable=import-error
@@ -26,7 +28,7 b' def cffi_get_next_item(arg, pos):'
     return container[0]
 
 
-def patches(text, bins):
+def patches(text: bytes, bins: List[bytes]) -> bytes:
     lgt = len(bins)
     all = []
     if not lgt:

mercurial/changegroup.py

@@ -105,6 +105,164 b' def writechunks(ui, chunks, filename, vf'
                 os.unlink(cleanup)
 
 
+def _dbg_ubdl_line(
+    ui,
+    indent,
+    key,
+    base_value=None,
+    percentage_base=None,
+    percentage_key=None,
+):
+    """Print one line of debug_unbundle_debug_info"""
+    line = b"DEBUG-UNBUNDLING: "
+    line += b' ' * (2 * indent)
+    key += b":"
+    padding = b''
+    if base_value is not None:
+        assert len(key) + 1 + (2 * indent) <= _KEY_PART_WIDTH
+        line += key.ljust(_KEY_PART_WIDTH - (2 * indent))
+        if isinstance(base_value, float):
+            line += b"%14.3f seconds" % base_value
+        else:
+            line += b"%10d" % base_value
+            padding = b'            '
+    else:
+        line += key
+
+    if percentage_base is not None:
+        line += padding
+        padding = b''
+        assert base_value is not None
+        percentage = base_value * 100 // percentage_base
+        if percentage_key is not None:
+            line += b" (%3d%% of %s)" % (
+                percentage,
+                percentage_key,
+            )
+        else:
+            line += b" (%3d%%)" % percentage
+
+    line += b'\n'
+    ui.write_err(line)
+
+
+def _sumf(items):
+    # python < 3.8 does not support a `start=0.0` argument to sum
+    # So we have to cheat a bit until we drop support for those version
+    if not items:
+        return 0.0
+    return sum(items)
+
+
+def display_unbundle_debug_info(ui, debug_info):
+    """display an unbundling report from debug information"""
+    cl_info = []
+    mn_info = []
+    fl_info = []
+    _dispatch = [
+        (b'CHANGELOG:', cl_info),
+        (b'MANIFESTLOG:', mn_info),
+        (b'FILELOG:', fl_info),
+    ]
+    for e in debug_info:
+        for prefix, info in _dispatch:
+            if e["target-revlog"].startswith(prefix):
+                info.append(e)
+                break
+        else:
+            assert False, 'unreachable'
+    each_info = [
+        (b'changelog', cl_info),
+        (b'manifests', mn_info),
+        (b'files', fl_info),
+    ]
+
+    # General Revision Countss
+    _dbg_ubdl_line(ui, 0, b'revisions', len(debug_info))
+    for key, info in each_info:
+        if not info:
+            continue
+        _dbg_ubdl_line(ui, 1, key, len(info), len(debug_info))
+
+    # General Time spent
+    all_durations = [e['duration'] for e in debug_info]
+    all_durations.sort()
+    total_duration = _sumf(all_durations)
+    _dbg_ubdl_line(ui, 0, b'total-time', total_duration)
+
+    for key, info in each_info:
+        if not info:
+            continue
+        durations = [e['duration'] for e in info]
+        durations.sort()
+        _dbg_ubdl_line(ui, 1, key, _sumf(durations), total_duration)
+
+    # Count and cache reuse per delta types
+    each_types = {}
+    for key, info in each_info:
+        each_types[key] = types = {
+            b'full': 0,
+            b'full-cached': 0,
+            b'snapshot': 0,
+            b'snapshot-cached': 0,
+            b'delta': 0,
+            b'delta-cached': 0,
+            b'unknown': 0,
+            b'unknown-cached': 0,
+        }
+        for e in info:
+            types[e['type']] += 1
+            if e['using-cached-base']:
+                types[e['type'] + b'-cached'] += 1
+
+    EXPECTED_TYPES = (b'full', b'snapshot', b'delta', b'unknown')
+    if debug_info:
+        _dbg_ubdl_line(ui, 0, b'type-count')
+    for key, info in each_info:
+        if info:
+            _dbg_ubdl_line(ui, 1, key)
+        t = each_types[key]
+        for tn in EXPECTED_TYPES:
+            if t[tn]:
+                tc = tn + b'-cached'
+                _dbg_ubdl_line(ui, 2, tn, t[tn])
+                _dbg_ubdl_line(ui, 3, b'cached', t[tc], t[tn])
+
+    # time perf delta types and reuse
+    each_type_time = {}
+    for key, info in each_info:
+        each_type_time[key] = t = {
+            b'full': [],
+            b'full-cached': [],
+            b'snapshot': [],
+            b'snapshot-cached': [],
+            b'delta': [],
+            b'delta-cached': [],
+            b'unknown': [],
+            b'unknown-cached': [],
+        }
+        for e in info:
+            t[e['type']].append(e['duration'])
+            if e['using-cached-base']:
+                t[e['type'] + b'-cached'].append(e['duration'])
+        for t_key, value in list(t.items()):
+            value.sort()
+            t[t_key] = _sumf(value)
+
+    if debug_info:
+        _dbg_ubdl_line(ui, 0, b'type-time')
+    for key, info in each_info:
+        if info:
+            _dbg_ubdl_line(ui, 1, key)
+        t = each_type_time[key]
+        td = total_duration  # to same space on next lines
+        for tn in EXPECTED_TYPES:
+            if t[tn]:
+                tc = tn + b'-cached'
+                _dbg_ubdl_line(ui, 2, tn, t[tn], td, b"total")
+                _dbg_ubdl_line(ui, 3, b'cached', t[tc], td, b"total")
+
+
 class cg1unpacker:
     """Unpacker for cg1 changegroup streams.
 
@@ -254,7 +412,16 b' class cg1unpacker:'
                     pos = next
             yield closechunk()
 
-    def _unpackmanifests(self, repo, revmap, trp, prog, addrevisioncb=None):
+    def _unpackmanifests(
+        self,
+        repo,
+        revmap,
+        trp,
+        prog,
+        addrevisioncb=None,
+        debug_info=None,
+        delta_base_reuse_policy=None,
+    ):
         self.callback = prog.increment
         # no need to check for empty manifest group here:
         # if the result of the merge of 1 and 2 is the same in 3 and 4,
@@ -263,7 +430,14 b' class cg1unpacker:'
         self.manifestheader()
         deltas = self.deltaiter()
         storage = repo.manifestlog.getstorage(b'')
-        storage.addgroup(deltas, revmap, trp, addrevisioncb=addrevisioncb)
+        storage.addgroup(
+            deltas,
+            revmap,
+            trp,
+            addrevisioncb=addrevisioncb,
+            debug_info=debug_info,
+            delta_base_reuse_policy=delta_base_reuse_policy,
+        )
         prog.complete()
         self.callback = None
 
@@ -276,6 +450,7 b' class cg1unpacker:'
         targetphase=phases.draft,
         expectedtotal=None,
         sidedata_categories=None,
+        delta_base_reuse_policy=None,
     ):
         """Add the changegroup returned by source.read() to this repo.
         srctype is a string like 'push', 'pull', or 'unbundle'.  url is
@@ -289,9 +464,19 b' class cg1unpacker:'
 
         `sidedata_categories` is an optional set of the remote's sidedata wanted
         categories.
+
+        `delta_base_reuse_policy` is an optional argument, when set to a value
+        it will control the way the delta contained into the bundle are reused
+        when applied in the revlog.
+
+        See `DELTA_BASE_REUSE_*` entry in mercurial.revlogutils.constants.
         """
         repo = repo.unfiltered()
 
+        debug_info = None
+        if repo.ui.configbool(b'debug', b'unbundling-stats'):
+            debug_info = []
+
         # Only useful if we're adding sidedata categories. If both peers have
         # the same categories, then we simply don't do anything.
         adding_sidedata = (
@@ -366,6 +551,8 b' class cg1unpacker:'
                 alwayscache=True,
                 addrevisioncb=onchangelog,
                 duplicaterevisioncb=ondupchangelog,
+                debug_info=debug_info,
+                delta_base_reuse_policy=delta_base_reuse_policy,
             ):
                 repo.ui.develwarn(
                     b'applied empty changelog from changegroup',
@@ -413,6 +600,8 b' class cg1unpacker:'
                 trp,
                 progress,
                 addrevisioncb=on_manifest_rev,
+                debug_info=debug_info,
+                delta_base_reuse_policy=delta_base_reuse_policy,
             )
 
             needfiles = {}
@@ -449,6 +638,8 b' class cg1unpacker:'
                 efiles,
                 needfiles,
                 addrevisioncb=on_filelog_rev,
+                debug_info=debug_info,
+                delta_base_reuse_policy=delta_base_reuse_policy,
             )
 
             if sidedata_helpers:
@@ -567,6 +758,8 b' class cg1unpacker:'
                     b'changegroup-runhooks-%020i' % clstart,
                     lambda tr: repo._afterlock(runhooks),
                 )
+            if debug_info is not None:
+                display_unbundle_debug_info(repo.ui, debug_info)
         finally:
             repo.ui.flush()
         # never return 0 here:
@@ -626,9 +819,24 b' class cg3unpacker(cg2unpacker):'
         protocol_flags = 0
         return node, p1, p2, deltabase, cs, flags, protocol_flags
 
-    def _unpackmanifests(self, repo, revmap, trp, prog, addrevisioncb=None):
+    def _unpackmanifests(
+        self,
+        repo,
+        revmap,
+        trp,
+        prog,
+        addrevisioncb=None,
+        debug_info=None,
+        delta_base_reuse_policy=None,
+    ):
         super(cg3unpacker, self)._unpackmanifests(
-            repo, revmap, trp, prog, addrevisioncb=addrevisioncb
+            repo,
+            revmap,
+            trp,
+            prog,
+            addrevisioncb=addrevisioncb,
+            debug_info=debug_info,
+            delta_base_reuse_policy=delta_base_reuse_policy,
         )
         for chunkdata in iter(self.filelogheader, {}):
             # If we get here, there are directory manifests in the changegroup
@@ -636,7 +844,12 b' class cg3unpacker(cg2unpacker):'
             repo.ui.debug(b"adding %s revisions\n" % d)
             deltas = self.deltaiter()
             if not repo.manifestlog.getstorage(d).addgroup(
-                deltas, revmap, trp, addrevisioncb=addrevisioncb
+                deltas,
+                revmap,
+                trp,
+                addrevisioncb=addrevisioncb,
+                debug_info=debug_info,
+                delta_base_reuse_policy=delta_base_reuse_policy,
             ):
                 raise error.Abort(_(b"received dir revlog group is empty"))
 
@@ -869,6 +1082,7 b' def deltagroup('
     fullclnodes=None,
     precomputedellipsis=None,
     sidedata_helpers=None,
+    debug_info=None,
 ):
     """Calculate deltas for a set of revisions.
 
@@ -978,6 +1192,7 b' def deltagroup('
         assumehaveparentrevisions=not ellipses,
         deltamode=deltamode,
         sidedata_helpers=sidedata_helpers,
+        debug_info=debug_info,
     )
 
     for i, revision in enumerate(revisions):
@@ -1003,6 +1218,187 b' def deltagroup('
         progress.complete()
 
 
+def make_debug_info():
+    """ "build a "new" debug_info dictionnary
+
+    That dictionnary can be used to gather information about the bundle process
+    """
+    return {
+        'revision-total': 0,
+        'revision-changelog': 0,
+        'revision-manifest': 0,
+        'revision-files': 0,
+        'file-count': 0,
+        'merge-total': 0,
+        'available-delta': 0,
+        'available-full': 0,
+        'delta-against-prev': 0,
+        'delta-full': 0,
+        'delta-against-p1': 0,
+        'denied-delta-candeltafn': 0,
+        'denied-base-not-available': 0,
+        'reused-storage-delta': 0,
+        'computed-delta': 0,
+    }
+
+
+def merge_debug_info(base, other):
+    """merge the debug information from <other> into <base>
+
+    This function can be used to gather lower level information into higher level ones.
+    """
+    for key in (
+        'revision-total',
+        'revision-changelog',
+        'revision-manifest',
+        'revision-files',
+        'merge-total',
+        'available-delta',
+        'available-full',
+        'delta-against-prev',
+        'delta-full',
+        'delta-against-p1',
+        'denied-delta-candeltafn',
+        'denied-base-not-available',
+        'reused-storage-delta',
+        'computed-delta',
+    ):
+        base[key] += other[key]
+
+
+_KEY_PART_WIDTH = 17
+
+
+def _dbg_bdl_line(
+    ui,
+    indent,
+    key,
+    base_value=None,
+    percentage_base=None,
+    percentage_key=None,
+    percentage_ref=None,
+    extra=None,
+):
+    """Print one line of debug_bundle_debug_info"""
+    line = b"DEBUG-BUNDLING: "
+    line += b' ' * (2 * indent)
+    key += b":"
+    if base_value is not None:
+        assert len(key) + 1 + (2 * indent) <= _KEY_PART_WIDTH
+        line += key.ljust(_KEY_PART_WIDTH - (2 * indent))
+        line += b"%10d" % base_value
+    else:
+        line += key
+
+    if percentage_base is not None:
+        assert base_value is not None
+        percentage = base_value * 100 // percentage_base
+        if percentage_key is not None:
+            line += b" (%d%% of %s %d)" % (
+                percentage,
+                percentage_key,
+                percentage_ref,
+            )
+        else:
+            line += b" (%d%%)" % percentage
+
+    if extra:
+        line += b" "
+        line += extra
+
+    line += b'\n'
+    ui.write_err(line)
+
+
+def display_bundling_debug_info(
+    ui,
+    debug_info,
+    cl_debug_info,
+    mn_debug_info,
+    fl_debug_info,
+):
+    """display debug information gathered during a bundling through `ui`"""
+    d = debug_info
+    c = cl_debug_info
+    m = mn_debug_info
+    f = fl_debug_info
+    all_info = [
+        (b"changelog", b"cl", c),
+        (b"manifests", b"mn", m),
+        (b"files", b"fl", f),
+    ]
+    _dbg_bdl_line(ui, 0, b'revisions', d['revision-total'])
+    _dbg_bdl_line(ui, 1, b'changelog', d['revision-changelog'])
+    _dbg_bdl_line(ui, 1, b'manifest', d['revision-manifest'])
+    extra = b'(for %d revlogs)' % d['file-count']
+    _dbg_bdl_line(ui, 1, b'files', d['revision-files'], extra=extra)
+    if d['merge-total']:
+        _dbg_bdl_line(ui, 1, b'merge', d['merge-total'], d['revision-total'])
+    for k, __, v in all_info:
+        if v['merge-total']:
+            _dbg_bdl_line(ui, 2, k, v['merge-total'], v['revision-total'])
+
+    _dbg_bdl_line(ui, 0, b'deltas')
+    _dbg_bdl_line(
+        ui,
+        1,
+        b'from-storage',
+        d['reused-storage-delta'],
+        percentage_base=d['available-delta'],
+        percentage_key=b"available",
+        percentage_ref=d['available-delta'],
+    )
+
+    if d['denied-delta-candeltafn']:
+        _dbg_bdl_line(ui, 2, b'denied-fn', d['denied-delta-candeltafn'])
+    for __, k, v in all_info:
+        if v['denied-delta-candeltafn']:
+            _dbg_bdl_line(ui, 3, k, v['denied-delta-candeltafn'])
+
+    if d['denied-base-not-available']:
+        _dbg_bdl_line(ui, 2, b'denied-nb', d['denied-base-not-available'])
+    for k, __, v in all_info:
+        if v['denied-base-not-available']:
+            _dbg_bdl_line(ui, 3, k, v['denied-base-not-available'])
+
+    if d['computed-delta']:
+        _dbg_bdl_line(ui, 1, b'computed', d['computed-delta'])
+
+    if d['available-full']:
+        _dbg_bdl_line(
+            ui,
+            2,
+            b'full',
+            d['delta-full'],
+            percentage_base=d['available-full'],
+            percentage_key=b"native",
+            percentage_ref=d['available-full'],
+        )
+    for k, __, v in all_info:
+        if v['available-full']:
+            _dbg_bdl_line(
+                ui,
+                3,
+                k,
+                v['delta-full'],
+                percentage_base=v['available-full'],
+                percentage_key=b"native",
+                percentage_ref=v['available-full'],
+            )
+
+    if d['delta-against-prev']:
+        _dbg_bdl_line(ui, 2, b'previous', d['delta-against-prev'])
+    for k, __, v in all_info:
+        if v['delta-against-prev']:
+            _dbg_bdl_line(ui, 3, k, v['delta-against-prev'])
+
+    if d['delta-against-p1']:
+        _dbg_bdl_line(ui, 2, b'parent-1', d['delta-against-prev'])
+    for k, __, v in all_info:
+        if v['delta-against-p1']:
+            _dbg_bdl_line(ui, 3, k, v['delta-against-p1'])
+
+
 class cgpacker:
     def __init__(
         self,
@@ -1086,13 +1482,21 b' class cgpacker:'
             self._verbosenote = lambda s: None
 
     def generate(
-        self, commonrevs, clnodes, fastpathlinkrev, source, changelog=True
+        self,
+        commonrevs,
+        clnodes,
+        fastpathlinkrev,
+        source,
+        changelog=True,
     ):
         """Yield a sequence of changegroup byte chunks.
         If changelog is False, changelog data won't be added to changegroup
         """
 
+        debug_info = None
         repo = self._repo
+        if repo.ui.configbool(b'debug', b'bundling-stats'):
+            debug_info = make_debug_info()
         cl = repo.changelog
 
         self._verbosenote(_(b'uncompressed size of bundle content:\n'))
@@ -1107,14 +1511,19 b' class cgpacker:'
                 # correctly advertise its sidedata categories directly.
                 remote_sidedata = repo._wanted_sidedata
             sidedata_helpers = sidedatamod.get_sidedata_helpers(
-                repo, remote_sidedata
+                repo,
+                remote_sidedata,
             )
 
+        cl_debug_info = None
+        if debug_info is not None:
+            cl_debug_info = make_debug_info()
         clstate, deltas = self._generatechangelog(
             cl,
             clnodes,
             generate=changelog,
             sidedata_helpers=sidedata_helpers,
+            debug_info=cl_debug_info,
         )
         for delta in deltas:
             for chunk in _revisiondeltatochunks(
@@ -1126,6 +1535,9 b' class cgpacker:'
         close = closechunk()
         size += len(close)
         yield closechunk()
+        if debug_info is not None:
+            merge_debug_info(debug_info, cl_debug_info)
+            debug_info['revision-changelog'] = cl_debug_info['revision-total']
 
         self._verbosenote(_(b'%8.i (changelog)\n') % size)
 
@@ -1133,6 +1545,9 b' class cgpacker:'
         manifests = clstate[b'manifests']
         changedfiles = clstate[b'changedfiles']
 
+        if debug_info is not None:
+            debug_info['file-count'] = len(changedfiles)
+
         # We need to make sure that the linkrev in the changegroup refers to
         # the first changeset that introduced the manifest or file revision.
         # The fastpath is usually safer than the slowpath, because the filelogs
@@ -1156,6 +1571,9 b' class cgpacker:'
         fnodes = {}  # needed file nodes
 
         size = 0
+        mn_debug_info = None
+        if debug_info is not None:
+            mn_debug_info = make_debug_info()
         it = self.generatemanifests(
             commonrevs,
             clrevorder,
@@ -1165,6 +1583,7 b' class cgpacker:'
             source,
             clstate[b'clrevtomanifestrev'],
             sidedata_helpers=sidedata_helpers,
+            debug_info=mn_debug_info,
         )
 
         for tree, deltas in it:
@@ -1185,6 +1604,9 b' class cgpacker:'
             close = closechunk()
             size += len(close)
             yield close
+        if debug_info is not None:
+            merge_debug_info(debug_info, mn_debug_info)
+            debug_info['revision-manifest'] = mn_debug_info['revision-total']
 
         self._verbosenote(_(b'%8.i (manifests)\n') % size)
         yield self._manifestsend
@@ -1199,6 +1621,9 b' class cgpacker:'
         manifests.clear()
         clrevs = {cl.rev(x) for x in clnodes}
 
+        fl_debug_info = None
+        if debug_info is not None:
+            fl_debug_info = make_debug_info()
         it = self.generatefiles(
             changedfiles,
             commonrevs,
@@ -1208,6 +1633,7 b' class cgpacker:'
             fnodes,
             clrevs,
             sidedata_helpers=sidedata_helpers,
+            debug_info=fl_debug_info,
         )
 
         for path, deltas in it:
@@ -1230,12 +1656,29 b' class cgpacker:'
             self._verbosenote(_(b'%8.i  %s\n') % (size, path))
 
         yield closechunk()
+        if debug_info is not None:
+            merge_debug_info(debug_info, fl_debug_info)
+            debug_info['revision-files'] = fl_debug_info['revision-total']
+
+        if debug_info is not None:
+            display_bundling_debug_info(
+                repo.ui,
+                debug_info,
+                cl_debug_info,
+                mn_debug_info,
+                fl_debug_info,
+            )
 
         if clnodes:
             repo.hook(b'outgoing', node=hex(clnodes[0]), source=source)
 
     def _generatechangelog(
-        self, cl, nodes, generate=True, sidedata_helpers=None
+        self,
+        cl,
+        nodes,
+        generate=True,
+        sidedata_helpers=None,
+        debug_info=None,
     ):
         """Generate data for changelog chunks.
 
@@ -1332,6 +1775,7 b' class cgpacker:'
             fullclnodes=self._fullclnodes,
             precomputedellipsis=self._precomputedellipsis,
             sidedata_helpers=sidedata_helpers,
+            debug_info=debug_info,
         )
 
         return state, gen
@@ -1346,6 +1790,7 b' class cgpacker:'
         source,
         clrevtolocalrev,
         sidedata_helpers=None,
+        debug_info=None,
     ):
         """Returns an iterator of changegroup chunks containing manifests.
 
@@ -1444,6 +1889,7 b' class cgpacker:'
                 fullclnodes=self._fullclnodes,
                 precomputedellipsis=self._precomputedellipsis,
                 sidedata_helpers=sidedata_helpers,
+                debug_info=debug_info,
             )
 
             if not self._oldmatcher.visitdir(store.tree[:-1]):
@@ -1483,6 +1929,7 b' class cgpacker:'
         fnodes,
         clrevs,
         sidedata_helpers=None,
+        debug_info=None,
     ):
         changedfiles = [
             f
@@ -1578,6 +2025,7 b' class cgpacker:'
                 fullclnodes=self._fullclnodes,
                 precomputedellipsis=self._precomputedellipsis,
                 sidedata_helpers=sidedata_helpers,
+                debug_info=debug_info,
             )
 
             yield fname, deltas
@@ -1867,7 +2315,12 b' def _changegroupinfo(repo, nodes, source'
 
 
 def makechangegroup(
-    repo, outgoing, version, source, fastpath=False, bundlecaps=None
+    repo,
+    outgoing,
+    version,
+    source,
+    fastpath=False,
+    bundlecaps=None,
 ):
     cgstream = makestream(
         repo,
@@ -1917,7 +2370,12 b' def makestream('
 
     repo.hook(b'preoutgoing', throw=True, source=source)
     _changegroupinfo(repo, csets, source)
-    return bundler.generate(commonrevs, csets, fastpathlinkrev, source)
+    return bundler.generate(
+        commonrevs,
+        csets,
+        fastpathlinkrev,
+        source,
+    )
 
 
 def _addchangegroupfiles(
@@ -1928,6 +2386,8 b' def _addchangegroupfiles('
     expectedfiles,
     needfiles,
     addrevisioncb=None,
+    debug_info=None,
+    delta_base_reuse_policy=None,
 ):
     revisions = 0
     files = 0
@@ -1948,6 +2408,8 b' def _addchangegroupfiles('
                 revmap,
                 trp,
                 addrevisioncb=addrevisioncb,
+                debug_info=debug_info,
+                delta_base_reuse_policy=delta_base_reuse_policy,
             )
             if not added:
                 raise error.Abort(_(b"received file revlog group is empty"))

mercurial/cmdutil.py

@@ -11,6 +11,15 b' import errno'
 import os
 import re
 
+from typing import (
+    Any,
+    AnyStr,
+    Dict,
+    Iterable,
+    Optional,
+    cast,
+)
+
 from .i18n import _
 from .node import (
     hex,
@@ -29,7 +38,6 b' from . import ('
     changelog,
     copies,
     crecord as crecordmod,
-    dirstateguard,
     encoding,
     error,
     formatter,
@@ -65,14 +73,10 b' from .revlogutils import ('
 )
 
 if pycompat.TYPE_CHECKING:
-    from typing import (
-        Any,
-        Dict,
+    from . import (
+        ui as uimod,
     )
 
-    for t in (Any, Dict):
-        assert t
-
 stringio = util.stringio
 
 # templates of common command options
@@ -269,13 +273,16 b' debugrevlogopts = ['
 _linebelow = b"^HG: ------------------------ >8 ------------------------$"
 
 
-def check_at_most_one_arg(opts, *args):
+def check_at_most_one_arg(
+    opts: Dict[AnyStr, Any],
+    *args: AnyStr,
+) -> Optional[AnyStr]:
     """abort if more than one of the arguments are in opts
 
     Returns the unique argument or None if none of them were specified.
     """
 
-    def to_display(name):
+    def to_display(name: AnyStr) -> bytes:
         return pycompat.sysbytes(name).replace(b'_', b'-')
 
     previous = None
@@ -290,7 +297,11 b' def check_at_most_one_arg(opts, *args):'
     return previous
 
 
-def check_incompatible_arguments(opts, first, others):
+def check_incompatible_arguments(
+    opts: Dict[AnyStr, Any],
+    first: AnyStr,
+    others: Iterable[AnyStr],
+) -> None:
     """abort if the first argument is given along with any of the others
 
     Unlike check_at_most_one_arg(), `others` are not mutually exclusive
@@ -300,7 +311,7 b' def check_incompatible_arguments(opts, f'
         check_at_most_one_arg(opts, first, other)
 
 
-def resolve_commit_options(ui, opts):
+def resolve_commit_options(ui: "uimod.ui", opts: Dict[str, Any]) -> bool:
     """modify commit options dict to handle related options
 
     The return value indicates that ``rewrite.update-timestamp`` is the reason
@@ -327,7 +338,7 b' def resolve_commit_options(ui, opts):'
     return datemaydiffer
 
 
-def check_note_size(opts):
+def check_note_size(opts: Dict[str, Any]) -> None:
     """make sure note is of valid format"""
 
     note = opts.get('note')
@@ -638,7 +649,7 b' def dorecord('
                         # already called within a `pendingchange`, However we
                         # are taking a shortcut here in order to be able to
                         # quickly deprecated the older API.
-                        with dirstate.parentchange():
+                        with dirstate.changing_parents(repo):
                             dirstate.update_file(
                                 realname,
                                 p1_tracked=True,
@@ -1115,12 +1126,12 b' def bailifchanged(repo, merge=True, hint'
         ctx.sub(s).bailifchanged(hint=hint)
 
 
-def logmessage(ui, opts):
+def logmessage(ui: "uimod.ui", opts: Dict[bytes, Any]) -> Optional[bytes]:
     """get the log message according to -m and -l option"""
 
     check_at_most_one_arg(opts, b'message', b'logfile')
 
-    message = opts.get(b'message')
+    message = cast(Optional[bytes], opts.get(b'message'))
     logfile = opts.get(b'logfile')
 
     if not message and logfile:
@@ -1465,7 +1476,7 b' def openrevlog(repo, cmd, file_, opts):'
     return openstorage(repo, cmd, file_, opts, returnrevlog=True)
 
 
-def copy(ui, repo, pats, opts, rename=False):
+def copy(ui, repo, pats, opts: Dict[bytes, Any], rename=False):
     check_incompatible_arguments(opts, b'forget', [b'dry_run'])
 
     # called with the repo lock held
@@ -1532,7 +1543,7 b' def copy(ui, repo, pats, opts, rename=Fa'
                 new_node = mem_ctx.commit()
 
                 if repo.dirstate.p1() == ctx.node():
-                    with repo.dirstate.parentchange():
+                    with repo.dirstate.changing_parents(repo):
                         scmutil.movedirstate(repo, repo[new_node])
                 replacements = {ctx.node(): [new_node]}
                 scmutil.cleanupnodes(
@@ -1625,7 +1636,7 b' def copy(ui, repo, pats, opts, rename=Fa'
             new_node = mem_ctx.commit()
 
             if repo.dirstate.p1() == ctx.node():
-                with repo.dirstate.parentchange():
+                with repo.dirstate.changing_parents(repo):
                     scmutil.movedirstate(repo, repo[new_node])
             replacements = {ctx.node(): [new_node]}
             scmutil.cleanupnodes(repo, replacements, b'copy', fixphase=True)
@@ -2778,7 +2789,7 b' def cat(ui, repo, ctx, matcher, basefm, '
                 basefm,
                 fntemplate,
                 subprefix,
-                **pycompat.strkwargs(opts)
+                **pycompat.strkwargs(opts),
             ):
                 err = 0
         except error.RepoLookupError:
@@ -2789,29 +2800,135 b' def cat(ui, repo, ctx, matcher, basefm, '
     return err
 
 
+class _AddRemoveContext:
+    """a small (hacky) context to deal with lazy opening of context
+
+    This is to be used in the `commit` function right below. This deals with
+    lazily open a `changing_files` context inside a `transaction` that span the
+    full commit operation.
+
+    We need :
+    - a `changing_files` context to wrap the dirstate change within the
+      "addremove" operation,
+    - a transaction to make sure these change are not written right after the
+      addremove, but when the commit operation succeed.
+
+    However it get complicated because:
+    - opening a transaction "this early" shuffle hooks order, especially the
+      `precommit` one happening after the `pretxtopen` one which I am not too
+      enthusiastic about.
+    - the `mq` extensions + the `record` extension stacks many layers of call
+      to implement `qrefresh --interactive` and this result with `mq` calling a
+      `strip` in the middle of this function. Which prevent the existence of
+      transaction wrapping all of its function code. (however, `qrefresh` never
+      call the `addremove` bits.
+    - the largefile extensions (and maybe other extensions?) wraps `addremove`
+      so slicing `addremove` in smaller bits is a complex endeavour.
+
+    So I eventually took a this shortcut that open the transaction if we
+    actually needs it, not disturbing much of the rest of the code.
+
+    It will result in some hooks order change for `hg commit --addremove`,
+    however it seems a corner case enough to ignore that for now (hopefully).
+
+    Notes that None of the above problems seems insurmountable, however I have
+    been fighting with this specific piece of code for a couple of day already
+    and I need a solution to keep moving forward on the bigger work around
+    `changing_files` context that is being introduced at the same time as this
+    hack.
+
+    Each problem seems to have a solution:
+    - the hook order issue could be solved by refactoring the many-layer stack
+      that currently composes a commit and calling them earlier,
+    - the mq issue could be solved by refactoring `mq` so that the final strip
+      is done after transaction closure. Be warned that the mq code is quite
+      antic however.
+    - large-file could be reworked in parallel of the `addremove` to be
+      friendlier to this.
+
+    However each of these tasks are too much a diversion right now. In addition
+    they will be much easier to undertake when the `changing_files` dust has
+    settled."""
+
+    def __init__(self, repo):
+        self._repo = repo
+        self._transaction = None
+        self._dirstate_context = None
+        self._state = None
+
+    def __enter__(self):
+        assert self._state is None
+        self._state = True
+        return self
+
+    def open_transaction(self):
+        """open a `transaction` and `changing_files` context
+
+        Call this when you know that change to the dirstate will be needed and
+        we need to open the transaction early
+
+        This will also open the dirstate `changing_files` context, so you should
+        call `close_dirstate_context` when the distate changes are done.
+        """
+        assert self._state is not None
+        if self._transaction is None:
+            self._transaction = self._repo.transaction(b'commit')
+            self._transaction.__enter__()
+        if self._dirstate_context is None:
+            self._dirstate_context = self._repo.dirstate.changing_files(
+                self._repo
+            )
+            self._dirstate_context.__enter__()
+
+    def close_dirstate_context(self):
+        """close the change_files if any
+
+        Call this after the (potential) `open_transaction` call to close the
+        (potential) changing_files context.
+        """
+        if self._dirstate_context is not None:
+            self._dirstate_context.__exit__(None, None, None)
+            self._dirstate_context = None
+
+    def __exit__(self, *args):
+        if self._dirstate_context is not None:
+            self._dirstate_context.__exit__(*args)
+        if self._transaction is not None:
+            self._transaction.__exit__(*args)
+
+
 def commit(ui, repo, commitfunc, pats, opts):
     '''commit the specified files or all outstanding changes'''
     date = opts.get(b'date')
     if date:
         opts[b'date'] = dateutil.parsedate(date)
-    message = logmessage(ui, opts)
-    matcher = scmutil.match(repo[None], pats, opts)
-
-    dsguard = None
-    # extract addremove carefully -- this function can be called from a command
-    # that doesn't support addremove
-    if opts.get(b'addremove'):
-        dsguard = dirstateguard.dirstateguard(repo, b'commit')
-    with dsguard or util.nullcontextmanager():
-        if dsguard:
-            relative = scmutil.anypats(pats, opts)
-            uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=relative)
-            if scmutil.addremove(repo, matcher, b"", uipathfn, opts) != 0:
-                raise error.Abort(
-                    _(b"failed to mark all new/missing files as added/removed")
+
+    with repo.wlock(), repo.lock():
+        message = logmessage(ui, opts)
+        matcher = scmutil.match(repo[None], pats, opts)
+
+        with _AddRemoveContext(repo) as c:
+            # extract addremove carefully -- this function can be called from a
+            # command that doesn't support addremove
+            if opts.get(b'addremove'):
+                relative = scmutil.anypats(pats, opts)
+                uipathfn = scmutil.getuipathfn(
+                    repo,
+                    legacyrelativevalue=relative,
                 )
-
-        return commitfunc(ui, repo, message, matcher, opts)
+                r = scmutil.addremove(
+                    repo,
+                    matcher,
+                    b"",
+                    uipathfn,
+                    opts,
+                    open_tr=c.open_transaction,
+                )
+                m = _(b"failed to mark all new/missing files as added/removed")
+                if r != 0:
+                    raise error.Abort(m)
+            c.close_dirstate_context()
+            return commitfunc(ui, repo, message, matcher, opts)
 
 
 def samefile(f, ctx1, ctx2):
@@ -2826,7 +2943,7 b' def samefile(f, ctx1, ctx2):'
         return f not in ctx2.manifest()
 
 
-def amend(ui, repo, old, extra, pats, opts):
+def amend(ui, repo, old, extra, pats, opts: Dict[str, Any]):
     # avoid cycle context -> subrepo -> cmdutil
     from . import context
 
@@ -2880,12 +2997,13 b' def amend(ui, repo, old, extra, pats, op'
         matcher = scmutil.match(wctx, pats, opts)
         relative = scmutil.anypats(pats, opts)
         uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=relative)
-        if opts.get(b'addremove') and scmutil.addremove(
-            repo, matcher, b"", uipathfn, opts
-        ):
-            raise error.Abort(
-                _(b"failed to mark all new/missing files as added/removed")
-            )
+        if opts.get(b'addremove'):
+            with repo.dirstate.changing_files(repo):
+                if scmutil.addremove(repo, matcher, b"", uipathfn, opts) != 0:
+                    m = _(
+                        b"failed to mark all new/missing files as added/removed"
+                    )
+                    raise error.Abort(m)
 
         # Check subrepos. This depends on in-place wctx._status update in
         # subrepo.precommit(). To minimize the risk of this hack, we do
@@ -3019,10 +3137,12 b' def amend(ui, repo, old, extra, pats, op'
         commitphase = None
         if opts.get(b'secret'):
             commitphase = phases.secret
+        elif opts.get(b'draft'):
+            commitphase = phases.draft
         newid = repo.commitctx(new)
         ms.reset()
 
-        with repo.dirstate.parentchange():
+        with repo.dirstate.changing_parents(repo):
             # Reroute the working copy parent to the new changeset
             repo.setparents(newid, repo.nullid)
 
@@ -3285,7 +3405,7 b' def revert(ui, repo, ctx, *pats, **opts)'
     names = {}
     uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=True)
 
-    with repo.wlock():
+    with repo.wlock(), repo.dirstate.changing_files(repo):
         ## filling of the `names` mapping
         # walk dirstate to fill `names`
 

mercurial/commands.py

@@ -13,6 +13,7 b' import sys'
 from .i18n import _
 from .node import (
     hex,
+    nullid,
     nullrev,
     short,
     wdirrev,
@@ -28,7 +29,6 b' from . import ('
     copies,
     debugcommands as debugcommandsmod,
     destutil,
-    dirstateguard,
     discovery,
     encoding,
     error,
@@ -252,10 +252,11 b' def add(ui, repo, *pats, **opts):'
     Returns 0 if all files are successfully added.
     """
 
-    m = scmutil.match(repo[None], pats, pycompat.byteskwargs(opts))
-    uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=True)
-    rejected = cmdutil.add(ui, repo, m, b"", uipathfn, False, **opts)
-    return rejected and 1 or 0
+    with repo.wlock(), repo.dirstate.changing_files(repo):
+        m = scmutil.match(repo[None], pats, pycompat.byteskwargs(opts))
+        uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=True)
+        rejected = cmdutil.add(ui, repo, m, b"", uipathfn, False, **opts)
+        return rejected and 1 or 0
 
 
 @command(
@@ -330,10 +331,11 b' def addremove(ui, repo, *pats, **opts):'
     opts = pycompat.byteskwargs(opts)
     if not opts.get(b'similarity'):
         opts[b'similarity'] = b'100'
-    matcher = scmutil.match(repo[None], pats, opts)
-    relative = scmutil.anypats(pats, opts)
-    uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=relative)
-    return scmutil.addremove(repo, matcher, b"", uipathfn, opts)
+    with repo.wlock(), repo.dirstate.changing_files(repo):
+        matcher = scmutil.match(repo[None], pats, opts)
+        relative = scmutil.anypats(pats, opts)
+        uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=relative)
+        return scmutil.addremove(repo, matcher, b"", uipathfn, opts)
 
 
 @command(
@@ -822,7 +824,7 b' def _dobackout(ui, repo, node=None, rev='
     bheads = repo.branchheads(branch)
     rctx = scmutil.revsingle(repo, hex(parent))
     if not opts.get(b'merge') and op1 != node:
-        with dirstateguard.dirstateguard(repo, b'backout'):
+        with repo.transaction(b"backout"):
             overrides = {(b'ui', b'forcemerge'): opts.get(b'tool', b'')}
             with ui.configoverride(overrides, b'backout'):
                 stats = mergemod.back_out(ctx, parent=repo[parent])
@@ -1635,7 +1637,7 b' def bundle(ui, repo, fname, *dests, **op'
         missing = set()
         excluded = set()
         for path in urlutil.get_push_paths(repo, ui, dests):
-            other = hg.peer(repo, opts, path.rawloc)
+            other = hg.peer(repo, opts, path)
             if revs is not None:
                 hex_revs = [repo[r].hex() for r in revs]
             else:
@@ -2008,6 +2010,7 b' def clone(ui, source, dest=None, **opts)'
         (b'', b'close-branch', None, _(b'mark a branch head as closed')),
         (b'', b'amend', None, _(b'amend the parent of the working directory')),
         (b's', b'secret', None, _(b'use the secret phase for committing')),
+        (b'', b'draft', None, _(b'use the draft phase for committing')),
         (b'e', b'edit', None, _(b'invoke editor on commit messages')),
         (
             b'',
@@ -2082,6 +2085,8 b' def commit(ui, repo, *pats, **opts):'
 
           hg commit --amend --date now
     """
+    cmdutil.check_at_most_one_arg(opts, 'draft', 'secret')
+    cmdutil.check_incompatible_arguments(opts, 'subrepos', ['amend'])
     with repo.wlock(), repo.lock():
         return _docommit(ui, repo, *pats, **opts)
 
@@ -2097,7 +2102,6 b' def _docommit(ui, repo, *pats, **opts):'
         return 1 if ret == 0 else ret
 
     if opts.get('subrepos'):
-        cmdutil.check_incompatible_arguments(opts, 'subrepos', ['amend'])
         # Let --subrepos on the command line override config setting.
         ui.setconfig(b'ui', b'commitsubrepos', True, b'commit')
 
@@ -2174,6 +2178,8 b' def _docommit(ui, repo, *pats, **opts):'
             overrides = {}
             if opts.get(b'secret'):
                 overrides[(b'phases', b'new-commit')] = b'secret'
+            elif opts.get(b'draft'):
+                overrides[(b'phases', b'new-commit')] = b'draft'
 
             baseui = repo.baseui
             with baseui.configoverride(overrides, b'commit'):
@@ -2491,7 +2497,19 b' def copy(ui, repo, *pats, **opts):'
     Returns 0 on success, 1 if errors are encountered.
     """
     opts = pycompat.byteskwargs(opts)
-    with repo.wlock():
+
+    context = repo.dirstate.changing_files
+    rev = opts.get(b'at_rev')
+    ctx = None
+    if rev:
+        ctx = logcmdutil.revsingle(repo, rev)
+        if ctx.rev() is not None:
+
+            def context(repo):
+                return util.nullcontextmanager()
+
+            opts[b'at_rev'] = ctx.rev()
+    with repo.wlock(), context(repo):
         return cmdutil.copy(ui, repo, pats, opts)
 
 
@@ -2960,19 +2978,20 b' def forget(ui, repo, *pats, **opts):'
     if not pats:
         raise error.InputError(_(b'no files specified'))
 
-    m = scmutil.match(repo[None], pats, opts)
-    dryrun, interactive = opts.get(b'dry_run'), opts.get(b'interactive')
-    uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=True)
-    rejected = cmdutil.forget(
-        ui,
-        repo,
-        m,
-        prefix=b"",
-        uipathfn=uipathfn,
-        explicitonly=False,
-        dryrun=dryrun,
-        interactive=interactive,
-    )[0]
+    with repo.wlock(), repo.dirstate.changing_files(repo):
+        m = scmutil.match(repo[None], pats, opts)
+        dryrun, interactive = opts.get(b'dry_run'), opts.get(b'interactive')
+        uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=True)
+        rejected = cmdutil.forget(
+            ui,
+            repo,
+            m,
+            prefix=b"",
+            uipathfn=uipathfn,
+            explicitonly=False,
+            dryrun=dryrun,
+            interactive=interactive,
+        )[0]
     return rejected and 1 or 0
 
 
@@ -3911,12 +3930,11 b' def identify('
     peer = None
     try:
         if source:
-            source, branches = urlutil.get_unique_pull_path(
-                b'identify', repo, ui, source
-            )
+            path = urlutil.get_unique_pull_path_obj(b'identify', ui, source)
             # only pass ui when no repo
-            peer = hg.peer(repo or ui, opts, source)
+            peer = hg.peer(repo or ui, opts, path)
             repo = peer.local()
+            branches = (path.branch, [])
             revs, checkout = hg.addbranchrevs(repo, peer, branches, None)
 
         fm = ui.formatter(b'identify', opts)
@@ -4229,12 +4247,10 b' def import_(ui, repo, patch1=None, *patc'
         if not opts.get(b'no_commit'):
             lock = repo.lock
             tr = lambda: repo.transaction(b'import')
-            dsguard = util.nullcontextmanager
         else:
             lock = util.nullcontextmanager
             tr = util.nullcontextmanager
-            dsguard = lambda: dirstateguard.dirstateguard(repo, b'import')
-        with lock(), tr(), dsguard():
+        with lock(), tr():
             parents = repo[None].parents()
             for patchurl in patches:
                 if patchurl == b'-':
@@ -4383,17 +4399,15 b' def incoming(ui, repo, source=b"default"'
     if opts.get(b'bookmarks'):
         srcs = urlutil.get_pull_paths(repo, ui, [source])
         for path in srcs:
-            source, branches = urlutil.parseurl(
-                path.rawloc, opts.get(b'branch')
-            )
-            other = hg.peer(repo, opts, source)
+            # XXX the "branches" options are not used. Should it be used?
+            other = hg.peer(repo, opts, path)
             try:
                 if b'bookmarks' not in other.listkeys(b'namespaces'):
                     ui.warn(_(b"remote doesn't support bookmarks\n"))
                     return 0
                 ui.pager(b'incoming')
                 ui.status(
-                    _(b'comparing with %s\n') % urlutil.hidepassword(source)
+                    _(b'comparing with %s\n') % urlutil.hidepassword(path.loc)
                 )
                 return bookmarks.incoming(
                     ui, repo, other, mode=path.bookmarks_mode
@@ -4426,7 +4440,7 b' def init(ui, dest=b".", **opts):'
     Returns 0 on success.
     """
     opts = pycompat.byteskwargs(opts)
-    path = urlutil.get_clone_path(ui, dest)[1]
+    path = urlutil.get_clone_path_obj(ui, dest)
     peer = hg.peer(ui, opts, path, create=True)
     peer.close()
 
@@ -5038,14 +5052,13 b' def outgoing(ui, repo, *dests, **opts):'
     opts = pycompat.byteskwargs(opts)
     if opts.get(b'bookmarks'):
         for path in urlutil.get_push_paths(repo, ui, dests):
-            dest = path.pushloc or path.loc
-            other = hg.peer(repo, opts, dest)
+            other = hg.peer(repo, opts, path)
             try:
                 if b'bookmarks' not in other.listkeys(b'namespaces'):
                     ui.warn(_(b"remote doesn't support bookmarks\n"))
                     return 0
                 ui.status(
-                    _(b'comparing with %s\n') % urlutil.hidepassword(dest)
+                    _(b'comparing with %s\n') % urlutil.hidepassword(path.loc)
                 )
                 ui.pager(b'outgoing')
                 return bookmarks.outgoing(ui, repo, other)
@@ -5434,12 +5447,12 b' def pull(ui, repo, *sources, **opts):'
         raise error.InputError(msg, hint=hint)
 
     for path in urlutil.get_pull_paths(repo, ui, sources):
-        source, branches = urlutil.parseurl(path.rawloc, opts.get(b'branch'))
-        ui.status(_(b'pulling from %s\n') % urlutil.hidepassword(source))
+        ui.status(_(b'pulling from %s\n') % urlutil.hidepassword(path.loc))
         ui.flush()
-        other = hg.peer(repo, opts, source)
+        other = hg.peer(repo, opts, path)
         update_conflict = None
         try:
+            branches = (path.branch, opts.get(b'branch', []))
             revs, checkout = hg.addbranchrevs(
                 repo, other, branches, opts.get(b'rev')
             )
@@ -5515,8 +5528,12 b' def pull(ui, repo, *sources, **opts):'
                     elif opts.get(b'branch'):
                         brev = opts[b'branch'][0]
                     else:
-                        brev = branches[0]
-                repo._subtoppath = source
+                        brev = path.branch
+
+                # XXX path: we are losing the `path` object here. Keeping it
+                # would be valuable. For example as a "variant" as we do
+                # for pushes.
+                repo._subtoppath = path.loc
                 try:
                     update_conflict = postincoming(
                         ui, repo, modheads, opts.get(b'update'), checkout, brev
@@ -5766,7 +5783,7 b' def push(ui, repo, *dests, **opts):'
     some_pushed = False
     result = 0
     for path in urlutil.get_push_paths(repo, ui, dests):
-        dest = path.pushloc or path.loc
+        dest = path.loc
         branches = (path.branch, opts.get(b'branch') or [])
         ui.status(_(b'pushing to %s\n') % urlutil.hidepassword(dest))
         revs, checkout = hg.addbranchrevs(
@@ -5940,12 +5957,13 b' def remove(ui, repo, *pats, **opts):'
     if not pats and not after:
         raise error.InputError(_(b'no files specified'))
 
-    m = scmutil.match(repo[None], pats, opts)
-    subrepos = opts.get(b'subrepos')
-    uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=True)
-    return cmdutil.remove(
-        ui, repo, m, b"", uipathfn, after, force, subrepos, dryrun=dryrun
-    )
+    with repo.wlock(), repo.dirstate.changing_files(repo):
+        m = scmutil.match(repo[None], pats, opts)
+        subrepos = opts.get(b'subrepos')
+        uipathfn = scmutil.getuipathfn(repo, legacyrelativevalue=True)
+        return cmdutil.remove(
+            ui, repo, m, b"", uipathfn, after, force, subrepos, dryrun=dryrun
+        )
 
 
 @command(
@@ -5994,7 +6012,18 b' def rename(ui, repo, *pats, **opts):'
     Returns 0 on success, 1 if errors are encountered.
     """
     opts = pycompat.byteskwargs(opts)
-    with repo.wlock():
+    context = repo.dirstate.changing_files
+    rev = opts.get(b'at_rev')
+    ctx = None
+    if rev:
+        ctx = logcmdutil.revsingle(repo, rev)
+        if ctx.rev() is not None:
+
+            def context(repo):
+                return util.nullcontextmanager()
+
+            opts[b'at_rev'] = ctx.rev()
+    with repo.wlock(), context(repo):
         return cmdutil.copy(ui, repo, pats, opts, rename=True)
 
 
@@ -6260,7 +6289,7 b' def resolve(ui, repo, *pats, **opts):'
         #
         # All this should eventually happens, but in the mean time, we use this
         # context manager slightly out of the context it should be.
-        with repo.dirstate.parentchange():
+        with repo.dirstate.changing_parents(repo):
             mergestatemod.recordupdates(repo, ms.actions(), branchmerge, None)
 
         if not didwork and pats:
@@ -7252,23 +7281,22 b' def summary(ui, repo, **opts):'
         # XXX We should actually skip this if no default is specified, instead
         # of passing "default" which will resolve as "./default/" if no default
         # path is defined.
-        source, branches = urlutil.get_unique_pull_path(
-            b'summary', repo, ui, b'default'
-        )
-        sbranch = branches[0]
+        path = urlutil.get_unique_pull_path_obj(b'summary', ui, b'default')
+        sbranch = path.branch
         try:
-            other = hg.peer(repo, {}, source)
+            other = hg.peer(repo, {}, path)
         except error.RepoError:
             if opts.get(b'remote'):
                 raise
-            return source, sbranch, None, None, None
+            return path.loc, sbranch, None, None, None
+        branches = (path.branch, [])
         revs, checkout = hg.addbranchrevs(repo, other, branches, None)
         if revs:
             revs = [other.lookup(rev) for rev in revs]
-        ui.debug(b'comparing with %s\n' % urlutil.hidepassword(source))
+        ui.debug(b'comparing with %s\n' % urlutil.hidepassword(path.loc))
         with repo.ui.silent():
             commoninc = discovery.findcommonincoming(repo, other, heads=revs)
-        return source, sbranch, other, commoninc, commoninc[1]
+        return path.loc, sbranch, other, commoninc, commoninc[1]
 
     if needsincoming:
         source, sbranch, sother, commoninc, incoming = getincoming()
@@ -7284,9 +7312,10 b' def summary(ui, repo, **opts):'
             d = b'default-push'
         elif b'default' in ui.paths:
             d = b'default'
+        path = None
         if d is not None:
             path = urlutil.get_unique_push_path(b'summary', repo, ui, d)
-            dest = path.pushloc or path.loc
+            dest = path.loc
             dbranch = path.branch
         else:
             dest = b'default'
@@ -7294,7 +7323,7 b' def summary(ui, repo, **opts):'
         revs, checkout = hg.addbranchrevs(repo, repo, (dbranch, []), None)
         if source != dest:
             try:
-                dother = hg.peer(repo, {}, dest)
+                dother = hg.peer(repo, {}, path if path is not None else dest)
             except error.RepoError:
                 if opts.get(b'remote'):
                     raise
@@ -7472,8 +7501,11 b' def tag(ui, repo, name1, *names, **opts)'
                 )
         node = logcmdutil.revsingle(repo, rev_).node()
 
+        # don't allow tagging the null rev or the working directory
         if node is None:
             raise error.InputError(_(b"cannot tag working directory"))
+        elif not opts.get(b'remove') and node == nullid:
+            raise error.InputError(_(b"cannot tag null revision"))
 
         if not message:
             # we don't translate commit messages
@@ -7494,13 +7526,6 b' def tag(ui, repo, name1, *names, **opts)'
             editform=editform, **pycompat.strkwargs(opts)
         )
 
-        # don't allow tagging the null rev
-        if (
-            not opts.get(b'remove')
-            and logcmdutil.revsingle(repo, rev_).rev() == nullrev
-        ):
-            raise error.InputError(_(b"cannot tag null revision"))
-
         tagsmod.tag(
             repo,
             names,

mercurial/configitems.py

@@ -588,6 +588,18 b' coreconfigitem('
     b'revlog.debug-delta',
     default=False,
 )
+# display extra information about the bundling process
+coreconfigitem(
+    b'debug',
+    b'bundling-stats',
+    default=False,
+)
+# display extra information about the unbundling process
+coreconfigitem(
+    b'debug',
+    b'unbundling-stats',
+    default=False,
+)
 coreconfigitem(
     b'defaults',
     b'.*',
@@ -734,6 +746,14 b' coreconfigitem('
     b'discovery.exchange-heads',
     default=True,
 )
+# If devel.debug.abort-update is True, then any merge with the working copy,
+# e.g. [hg update], will be aborted after figuring out what needs to be done,
+# but before spawning the parallel worker
+coreconfigitem(
+    b'devel',
+    b'debug.abort-update',
+    default=False,
+)
 # If discovery.grow-sample is False, the sample size used in set discovery will
 # not be increased through the process
 coreconfigitem(
@@ -911,6 +931,13 b' coreconfigitem('
     b'changegroup4',
     default=False,
 )
+
+# might remove rank configuration once the computation has no impact
+coreconfigitem(
+    b'experimental',
+    b'changelog-v2.compute-rank',
+    default=True,
+)
 coreconfigitem(
     b'experimental',
     b'cleanup-as-archived',
@@ -1774,6 +1801,13 b' coreconfigitem('
 )
 coreconfigitem(
     b'merge-tools',
+    br'.*\.regappend$',
+    default=b"",
+    generic=True,
+    priority=-1,
+)
+coreconfigitem(
+    b'merge-tools',
     br'.*\.symlink$',
     default=False,
     generic=True,
@@ -2023,6 +2057,11 b' coreconfigitem('
 )
 coreconfigitem(
     b'storage',
+    b'revlog.delta-parent-search.candidate-group-chunk-size',
+    default=10,
+)
+coreconfigitem(
+    b'storage',
     b'revlog.issue6528.fix-incoming',
     default=True,
 )
@@ -2044,6 +2083,7 b' coreconfigitem('
     b'revlog.reuse-external-delta',
     default=True,
 )
+# This option is True unless `format.generaldelta` is set.
 coreconfigitem(
     b'storage',
     b'revlog.reuse-external-delta-parent',
@@ -2123,7 +2163,7 b' coreconfigitem('
 coreconfigitem(
     b'server',
     b'pullbundle',
-    default=False,
+    default=True,
 )
 coreconfigitem(
     b'server',

mercurial/context.py

@@ -1595,7 +1595,7 b' class workingctx(committablectx):'
         if p2node is None:
             p2node = self._repo.nodeconstants.nullid
         dirstate = self._repo.dirstate
-        with dirstate.parentchange():
+        with dirstate.changing_parents(self._repo):
             copies = dirstate.setparents(p1node, p2node)
             pctx = self._repo[p1node]
             if copies:
@@ -1854,47 +1854,42 b' class workingctx(committablectx):'
 
     def _poststatusfixup(self, status, fixup):
         """update dirstate for files that are actually clean"""
+        dirstate = self._repo.dirstate
         poststatus = self._repo.postdsstatus()
-        if fixup or poststatus or self._repo.dirstate._dirty:
+        if fixup:
+            if dirstate.is_changing_parents:
+                normal = lambda f, pfd: dirstate.update_file(
+                    f,
+                    p1_tracked=True,
+                    wc_tracked=True,
+                )
+            else:
+                normal = dirstate.set_clean
+            for f, pdf in fixup:
+                normal(f, pdf)
+        if poststatus or self._repo.dirstate._dirty:
             try:
-                oldid = self._repo.dirstate.identity()
-
                 # updating the dirstate is optional
                 # so we don't wait on the lock
                 # wlock can invalidate the dirstate, so cache normal _after_
                 # taking the lock
+                pre_dirty = dirstate._dirty
                 with self._repo.wlock(False):
-                    dirstate = self._repo.dirstate
-                    if dirstate.identity() == oldid:
-                        if fixup:
-                            if dirstate.pendingparentchange():
-                                normal = lambda f, pfd: dirstate.update_file(
-                                    f, p1_tracked=True, wc_tracked=True
-                                )
-                            else:
-                                normal = dirstate.set_clean
-                            for f, pdf in fixup:
-                                normal(f, pdf)
-                            # write changes out explicitly, because nesting
-                            # wlock at runtime may prevent 'wlock.release()'
-                            # after this block from doing so for subsequent
-                            # changing files
-                            tr = self._repo.currenttransaction()
-                            self._repo.dirstate.write(tr)
-
-                        if poststatus:
-                            for ps in poststatus:
-                                ps(self, status)
-                    else:
-                        # in this case, writing changes out breaks
-                        # consistency, because .hg/dirstate was
-                        # already changed simultaneously after last
-                        # caching (see also issue5584 for detail)
-                        self._repo.ui.debug(
-                            b'skip updating dirstate: identity mismatch\n'
-                        )
+                    assert self._repo.dirstate is dirstate
+                    post_dirty = dirstate._dirty
+                    if post_dirty:
+                        tr = self._repo.currenttransaction()
+                        dirstate.write(tr)
+                    elif pre_dirty:
+                        # the wlock grabbing detected that dirtate changes
+                        # needed to be dropped
+                        m = b'skip updating dirstate: identity mismatch\n'
+                        self._repo.ui.debug(m)
+                    if poststatus:
+                        for ps in poststatus:
+                            ps(self, status)
             except error.LockError:
-                pass
+                dirstate.invalidate()
             finally:
                 # Even if the wlock couldn't be grabbed, clear out the list.
                 self._repo.clearpostdsstatus()
@@ -1904,25 +1899,27 b' class workingctx(committablectx):'
         subrepos = []
         if b'.hgsub' in self:
             subrepos = sorted(self.substate)
-        cmp, s, mtime_boundary = self._repo.dirstate.status(
-            match, subrepos, ignored=ignored, clean=clean, unknown=unknown
-        )
-
-        # check for any possibly clean files
-        fixup = []
-        if cmp:
-            modified2, deleted2, clean_set, fixup = self._checklookup(
-                cmp, mtime_boundary
+        dirstate = self._repo.dirstate
+        with dirstate.running_status(self._repo):
+            cmp, s, mtime_boundary = dirstate.status(
+                match, subrepos, ignored=ignored, clean=clean, unknown=unknown
             )
-            s.modified.extend(modified2)
-            s.deleted.extend(deleted2)
-
-            if clean_set and clean:
-                s.clean.extend(clean_set)
-            if fixup and clean:
-                s.clean.extend((f for f, _ in fixup))
-
-        self._poststatusfixup(s, fixup)
+
+            # check for any possibly clean files
+            fixup = []
+            if cmp:
+                modified2, deleted2, clean_set, fixup = self._checklookup(
+                    cmp, mtime_boundary
+                )
+                s.modified.extend(modified2)
+                s.deleted.extend(deleted2)
+
+                if clean_set and clean:
+                    s.clean.extend(clean_set)
+                if fixup and clean:
+                    s.clean.extend((f for f, _ in fixup))
+
+            self._poststatusfixup(s, fixup)
 
         if match.always():
             # cache for performance
@@ -2050,7 +2047,7 b' class workingctx(committablectx):'
         return sorted(f for f in ds.matches(match) if ds.get_entry(f).tracked)
 
     def markcommitted(self, node):
-        with self._repo.dirstate.parentchange():
+        with self._repo.dirstate.changing_parents(self._repo):
             for f in self.modified() + self.added():
                 self._repo.dirstate.update_file(
                     f, p1_tracked=True, wc_tracked=True

mercurial/debugcommands.py

@@ -21,7 +21,6 b' import re'
 import socket
 import ssl
 import stat
-import string
 import subprocess
 import sys
 import time
@@ -73,7 +72,6 b' from . import ('
     repoview,
     requirements,
     revlog,
-    revlogutils,
     revset,
     revsetlang,
     scmutil,
@@ -89,6 +87,7 b' from . import ('
     upgrade,
     url as urlmod,
     util,
+    verify,
     vfs as vfsmod,
     wireprotoframing,
     wireprotoserver,
@@ -556,15 +555,9 b' def debugchangedfiles(ui, repo, rev, **o'
 @command(b'debugcheckstate', [], b'')
 def debugcheckstate(ui, repo):
     """validate the correctness of the current dirstate"""
-    parent1, parent2 = repo.dirstate.parents()
-    m1 = repo[parent1].manifest()
-    m2 = repo[parent2].manifest()
-    errors = 0
-    for err in repo.dirstate.verify(m1, m2):
-        ui.warn(err[0] % err[1:])
-        errors += 1
+    errors = verify.verifier(repo)._verify_dirstate()
     if errors:
-        errstr = _(b".hg/dirstate inconsistent with current parent's manifest")
+        errstr = _(b"dirstate inconsistent with current parent's manifest")
         raise error.Abort(errstr)
 
 
@@ -990,17 +983,29 b' def debugdeltachain(ui, repo, file_=None'
 
 @command(
     b'debug-delta-find',
-    cmdutil.debugrevlogopts + cmdutil.formatteropts,
+    cmdutil.debugrevlogopts
+    + cmdutil.formatteropts
+    + [
+        (
+            b'',
+            b'source',
+            b'full',
+            _(b'input data feed to the process (full, storage, p1, p2, prev)'),
+        ),
+    ],
     _(b'-c|-m|FILE REV'),
     optionalrepo=True,
 )
-def debugdeltafind(ui, repo, arg_1, arg_2=None, **opts):
+def debugdeltafind(ui, repo, arg_1, arg_2=None, source=b'full', **opts):
     """display the computation to get to a valid delta for storing REV
 
     This command will replay the process used to find the "best" delta to store
     a revision and display information about all the steps used to get to that
     result.
 
+    By default, the process is fed with a the full-text for the revision. This
+    can be controlled with the --source flag.
+
     The revision use the revision number of the target storage (not changelog
     revision number).
 
@@ -1017,34 +1022,22 b' def debugdeltafind(ui, repo, arg_1, arg_'
     rev = int(rev)
 
     revlog = cmdutil.openrevlog(repo, b'debugdeltachain', file_, opts)
-
-    deltacomputer = deltautil.deltacomputer(
-        revlog,
-        write_debug=ui.write,
-        debug_search=not ui.quiet,
-    )
-
-    node = revlog.node(rev)
     p1r, p2r = revlog.parentrevs(rev)
-    p1 = revlog.node(p1r)
-    p2 = revlog.node(p2r)
-    btext = [revlog.revision(rev)]
-    textlen = len(btext[0])
-    cachedelta = None
-    flags = revlog.flags(rev)
-
-    revinfo = revlogutils.revisioninfo(
-        node,
-        p1,
-        p2,
-        btext,
-        textlen,
-        cachedelta,
-        flags,
-    )
-
-    fh = revlog._datafp()
-    deltacomputer.finddeltainfo(revinfo, fh, target_rev=rev)
+
+    if source == b'full':
+        base_rev = nullrev
+    elif source == b'storage':
+        base_rev = revlog.deltaparent(rev)
+    elif source == b'p1':
+        base_rev = p1r
+    elif source == b'p2':
+        base_rev = p2r
+    elif source == b'prev':
+        base_rev = rev - 1
+    else:
+        raise error.InputError(b"invalid --source value: %s" % source)
+
+    revlog_debug.debug_delta_find(ui, revlog, rev, base_rev=base_rev)
 
 
 @command(
@@ -1236,12 +1229,12 b' def debugdiscovery(ui, repo, remoteurl=b'
     random.seed(int(opts[b'seed']))
 
     if not remote_revs:
-
-        remoteurl, branches = urlutil.get_unique_pull_path(
-            b'debugdiscovery', repo, ui, remoteurl
+        path = urlutil.get_unique_pull_path_obj(
+            b'debugdiscovery', ui, remoteurl
         )
-        remote = hg.peer(repo, opts, remoteurl)
-        ui.status(_(b'comparing with %s\n') % urlutil.hidepassword(remoteurl))
+        branches = (path.branch, [])
+        remote = hg.peer(repo, opts, path)
+        ui.status(_(b'comparing with %s\n') % urlutil.hidepassword(path.loc))
     else:
         branches = (None, [])
         remote_filtered_revs = logcmdutil.revrange(
@@ -3135,6 +3128,9 b' def debugrebuilddirstate(ui, repo, rev, '
     """
     ctx = scmutil.revsingle(repo, rev)
     with repo.wlock():
+        if repo.currenttransaction() is not None:
+            msg = b'rebuild the dirstate outside of a transaction'
+            raise error.ProgrammingError(msg)
         dirstate = repo.dirstate
         changedfiles = None
         # See command doc for what minimal does.
@@ -3146,7 +3142,8 b' def debugrebuilddirstate(ui, repo, rev, '
             dsnotadded = {f for f in dsonly if not dirstate.get_entry(f).added}
             changedfiles = manifestonly | dsnotadded
 
-        dirstate.rebuild(ctx.node(), ctx.manifest(), changedfiles)
+        with dirstate.changing_parents(repo):
+            dirstate.rebuild(ctx.node(), ctx.manifest(), changedfiles)
 
 
 @command(
@@ -3207,348 +3204,10 b' def debugrevlog(ui, repo, file_=None, **'
     r = cmdutil.openrevlog(repo, b'debugrevlog', file_, opts)
 
     if opts.get(b"dump"):
-        numrevs = len(r)
-        ui.write(
-            (
-                b"# rev p1rev p2rev start   end deltastart base   p1   p2"
-                b" rawsize totalsize compression heads chainlen\n"
-            )
-        )
-        ts = 0
-        heads = set()
-
-        for rev in range(numrevs):
-            dbase = r.deltaparent(rev)
-            if dbase == -1:
-                dbase = rev
-            cbase = r.chainbase(rev)
-            clen = r.chainlen(rev)
-            p1, p2 = r.parentrevs(rev)
-            rs = r.rawsize(rev)
-            ts = ts + rs
-            heads -= set(r.parentrevs(rev))
-            heads.add(rev)
-            try:
-                compression = ts / r.end(rev)
-            except ZeroDivisionError:
-                compression = 0
-            ui.write(
-                b"%5d %5d %5d %5d %5d %10d %4d %4d %4d %7d %9d "
-                b"%11d %5d %8d\n"
-                % (
-                    rev,
-                    p1,
-                    p2,
-                    r.start(rev),
-                    r.end(rev),
-                    r.start(dbase),
-                    r.start(cbase),
-                    r.start(p1),
-                    r.start(p2),
-                    rs,
-                    ts,
-                    compression,
-                    len(heads),
-                    clen,
-                )
-            )
-        return 0
-
-    format = r._format_version
-    v = r._format_flags
-    flags = []
-    gdelta = False
-    if v & revlog.FLAG_INLINE_DATA:
-        flags.append(b'inline')
-    if v & revlog.FLAG_GENERALDELTA:
-        gdelta = True
-        flags.append(b'generaldelta')
-    if not flags:
-        flags = [b'(none)']
-
-    ### tracks merge vs single parent
-    nummerges = 0
-
-    ### tracks ways the "delta" are build
-    # nodelta
-    numempty = 0
-    numemptytext = 0
-    numemptydelta = 0
-    # full file content
-    numfull = 0
-    # intermediate snapshot against a prior snapshot
-    numsemi = 0
-    # snapshot count per depth
-    numsnapdepth = collections.defaultdict(lambda: 0)
-    # delta against previous revision
-    numprev = 0
-    # delta against first or second parent (not prev)
-    nump1 = 0
-    nump2 = 0
-    # delta against neither prev nor parents
-    numother = 0
-    # delta against prev that are also first or second parent
-    # (details of `numprev`)
-    nump1prev = 0
-    nump2prev = 0
-
-    # data about delta chain of each revs
-    chainlengths = []
-    chainbases = []
-    chainspans = []
-
-    # data about each revision
-    datasize = [None, 0, 0]
-    fullsize = [None, 0, 0]
-    semisize = [None, 0, 0]
-    # snapshot count per depth
-    snapsizedepth = collections.defaultdict(lambda: [None, 0, 0])
-    deltasize = [None, 0, 0]
-    chunktypecounts = {}
-    chunktypesizes = {}
-
-    def addsize(size, l):
-        if l[0] is None or size < l[0]:
-            l[0] = size
-        if size > l[1]:
-            l[1] = size
-        l[2] += size
-
-    numrevs = len(r)
-    for rev in range(numrevs):
-        p1, p2 = r.parentrevs(rev)
-        delta = r.deltaparent(rev)
-        if format > 0:
-            addsize(r.rawsize(rev), datasize)
-        if p2 != nullrev:
-            nummerges += 1
-        size = r.length(rev)
-        if delta == nullrev:
-            chainlengths.append(0)
-            chainbases.append(r.start(rev))
-            chainspans.append(size)
-            if size == 0:
-                numempty += 1
-                numemptytext += 1
-            else:
-                numfull += 1
-                numsnapdepth[0] += 1
-                addsize(size, fullsize)
-                addsize(size, snapsizedepth[0])
-        else:
-            chainlengths.append(chainlengths[delta] + 1)
-            baseaddr = chainbases[delta]
-            revaddr = r.start(rev)
-            chainbases.append(baseaddr)
-            chainspans.append((revaddr - baseaddr) + size)
-            if size == 0:
-                numempty += 1
-                numemptydelta += 1
-            elif r.issnapshot(rev):
-                addsize(size, semisize)
-                numsemi += 1
-                depth = r.snapshotdepth(rev)
-                numsnapdepth[depth] += 1
-                addsize(size, snapsizedepth[depth])
-            else:
-                addsize(size, deltasize)
-                if delta == rev - 1:
-                    numprev += 1
-                    if delta == p1:
-                        nump1prev += 1
-                    elif delta == p2:
-                        nump2prev += 1
-                elif delta == p1:
-                    nump1 += 1
-                elif delta == p2:
-                    nump2 += 1
-                elif delta != nullrev:
-                    numother += 1
-
-        # Obtain data on the raw chunks in the revlog.
-        if util.safehasattr(r, b'_getsegmentforrevs'):
-            segment = r._getsegmentforrevs(rev, rev)[1]
-        else:
-            segment = r._revlog._getsegmentforrevs(rev, rev)[1]
-        if segment:
-            chunktype = bytes(segment[0:1])
-        else:
-            chunktype = b'empty'
-
-        if chunktype not in chunktypecounts:
-            chunktypecounts[chunktype] = 0
-            chunktypesizes[chunktype] = 0
-
-        chunktypecounts[chunktype] += 1
-        chunktypesizes[chunktype] += size
-
-    # Adjust size min value for empty cases
-    for size in (datasize, fullsize, semisize, deltasize):
-        if size[0] is None:
-            size[0] = 0
-
-    numdeltas = numrevs - numfull - numempty - numsemi
-    numoprev = numprev - nump1prev - nump2prev
-    totalrawsize = datasize[2]
-    datasize[2] /= numrevs
-    fulltotal = fullsize[2]
-    if numfull == 0:
-        fullsize[2] = 0
+        revlog_debug.dump(ui, r)
     else:
-        fullsize[2] /= numfull
-    semitotal = semisize[2]
-    snaptotal = {}
-    if numsemi > 0:
-        semisize[2] /= numsemi
-    for depth in snapsizedepth:
-        snaptotal[depth] = snapsizedepth[depth][2]
-        snapsizedepth[depth][2] /= numsnapdepth[depth]
-
-    deltatotal = deltasize[2]
-    if numdeltas > 0:
-        deltasize[2] /= numdeltas
-    totalsize = fulltotal + semitotal + deltatotal
-    avgchainlen = sum(chainlengths) / numrevs
-    maxchainlen = max(chainlengths)
-    maxchainspan = max(chainspans)
-    compratio = 1
-    if totalsize:
-        compratio = totalrawsize / totalsize
-
-    basedfmtstr = b'%%%dd\n'
-    basepcfmtstr = b'%%%dd %s(%%5.2f%%%%)\n'
-
-    def dfmtstr(max):
-        return basedfmtstr % len(str(max))
-
-    def pcfmtstr(max, padding=0):
-        return basepcfmtstr % (len(str(max)), b' ' * padding)
-
-    def pcfmt(value, total):
-        if total:
-            return (value, 100 * float(value) / total)
-        else:
-            return value, 100.0
-
-    ui.writenoi18n(b'format : %d\n' % format)
-    ui.writenoi18n(b'flags  : %s\n' % b', '.join(flags))
-
-    ui.write(b'\n')
-    fmt = pcfmtstr(totalsize)
-    fmt2 = dfmtstr(totalsize)
-    ui.writenoi18n(b'revisions     : ' + fmt2 % numrevs)
-    ui.writenoi18n(b'    merges    : ' + fmt % pcfmt(nummerges, numrevs))
-    ui.writenoi18n(
-        b'    normal    : ' + fmt % pcfmt(numrevs - nummerges, numrevs)
-    )
-    ui.writenoi18n(b'revisions     : ' + fmt2 % numrevs)
-    ui.writenoi18n(b'    empty     : ' + fmt % pcfmt(numempty, numrevs))
-    ui.writenoi18n(
-        b'                   text  : '
-        + fmt % pcfmt(numemptytext, numemptytext + numemptydelta)
-    )
-    ui.writenoi18n(
-        b'                   delta : '
-        + fmt % pcfmt(numemptydelta, numemptytext + numemptydelta)
-    )
-    ui.writenoi18n(
-        b'    snapshot  : ' + fmt % pcfmt(numfull + numsemi, numrevs)
-    )
-    for depth in sorted(numsnapdepth):
-        ui.write(
-            (b'      lvl-%-3d :       ' % depth)
-            + fmt % pcfmt(numsnapdepth[depth], numrevs)
-        )
-    ui.writenoi18n(b'    deltas    : ' + fmt % pcfmt(numdeltas, numrevs))
-    ui.writenoi18n(b'revision size : ' + fmt2 % totalsize)
-    ui.writenoi18n(
-        b'    snapshot  : ' + fmt % pcfmt(fulltotal + semitotal, totalsize)
-    )
-    for depth in sorted(numsnapdepth):
-        ui.write(
-            (b'      lvl-%-3d :       ' % depth)
-            + fmt % pcfmt(snaptotal[depth], totalsize)
-        )
-    ui.writenoi18n(b'    deltas    : ' + fmt % pcfmt(deltatotal, totalsize))
-
-    def fmtchunktype(chunktype):
-        if chunktype == b'empty':
-            return b'    %s     : ' % chunktype
-        elif chunktype in pycompat.bytestr(string.ascii_letters):
-            return b'    0x%s (%s)  : ' % (hex(chunktype), chunktype)
-        else:
-            return b'    0x%s      : ' % hex(chunktype)
-
-    ui.write(b'\n')
-    ui.writenoi18n(b'chunks        : ' + fmt2 % numrevs)
-    for chunktype in sorted(chunktypecounts):
-        ui.write(fmtchunktype(chunktype))
-        ui.write(fmt % pcfmt(chunktypecounts[chunktype], numrevs))
-    ui.writenoi18n(b'chunks size   : ' + fmt2 % totalsize)
-    for chunktype in sorted(chunktypecounts):
-        ui.write(fmtchunktype(chunktype))
-        ui.write(fmt % pcfmt(chunktypesizes[chunktype], totalsize))
-
-    ui.write(b'\n')
-    fmt = dfmtstr(max(avgchainlen, maxchainlen, maxchainspan, compratio))
-    ui.writenoi18n(b'avg chain length  : ' + fmt % avgchainlen)
-    ui.writenoi18n(b'max chain length  : ' + fmt % maxchainlen)
-    ui.writenoi18n(b'max chain reach   : ' + fmt % maxchainspan)
-    ui.writenoi18n(b'compression ratio : ' + fmt % compratio)
-
-    if format > 0:
-        ui.write(b'\n')
-        ui.writenoi18n(
-            b'uncompressed data size (min/max/avg) : %d / %d / %d\n'
-            % tuple(datasize)
-        )
-    ui.writenoi18n(
-        b'full revision size (min/max/avg)     : %d / %d / %d\n'
-        % tuple(fullsize)
-    )
-    ui.writenoi18n(
-        b'inter-snapshot size (min/max/avg)    : %d / %d / %d\n'
-        % tuple(semisize)
-    )
-    for depth in sorted(snapsizedepth):
-        if depth == 0:
-            continue
-        ui.writenoi18n(
-            b'    level-%-3d (min/max/avg)          : %d / %d / %d\n'
-            % ((depth,) + tuple(snapsizedepth[depth]))
-        )
-    ui.writenoi18n(
-        b'delta size (min/max/avg)             : %d / %d / %d\n'
-        % tuple(deltasize)
-    )
-
-    if numdeltas > 0:
-        ui.write(b'\n')
-        fmt = pcfmtstr(numdeltas)
-        fmt2 = pcfmtstr(numdeltas, 4)
-        ui.writenoi18n(
-            b'deltas against prev  : ' + fmt % pcfmt(numprev, numdeltas)
-        )
-        if numprev > 0:
-            ui.writenoi18n(
-                b'    where prev = p1  : ' + fmt2 % pcfmt(nump1prev, numprev)
-            )
-            ui.writenoi18n(
-                b'    where prev = p2  : ' + fmt2 % pcfmt(nump2prev, numprev)
-            )
-            ui.writenoi18n(
-                b'    other            : ' + fmt2 % pcfmt(numoprev, numprev)
-            )
-        if gdelta:
-            ui.writenoi18n(
-                b'deltas against p1    : ' + fmt % pcfmt(nump1, numdeltas)
-            )
-            ui.writenoi18n(
-                b'deltas against p2    : ' + fmt % pcfmt(nump2, numdeltas)
-            )
-            ui.writenoi18n(
-                b'deltas against other : ' + fmt % pcfmt(numother, numdeltas)
-            )
+        revlog_debug.debug_revlog(ui, r)
+    return 0
 
 
 @command(
@@ -3935,10 +3594,8 b' def debugssl(ui, repo, source=None, **op'
             )
         source = b"default"
 
-    source, branches = urlutil.get_unique_pull_path(
-        b'debugssl', repo, ui, source
-    )
-    url = urlutil.url(source)
+    path = urlutil.get_unique_pull_path_obj(b'debugssl', ui, source)
+    url = path.url
 
     defaultport = {b'https': 443, b'ssh': 22}
     if url.scheme in defaultport:
@@ -4049,20 +3706,19 b' def debugbackupbundle(ui, repo, *pats, *'
     for backup in backups:
         # Much of this is copied from the hg incoming logic
         source = os.path.relpath(backup, encoding.getcwd())
-        source, branches = urlutil.get_unique_pull_path(
+        path = urlutil.get_unique_pull_path_obj(
             b'debugbackupbundle',
-            repo,
             ui,
             source,
-            default_branches=opts.get(b'branch'),
         )
         try:
-            other = hg.peer(repo, opts, source)
+            other = hg.peer(repo, opts, path)
         except error.LookupError as ex:
-            msg = _(b"\nwarning: unable to open bundle %s") % source
+            msg = _(b"\nwarning: unable to open bundle %s") % path.loc
             hint = _(b"\n(missing parent rev %s)\n") % short(ex.name)
             ui.warn(msg, hint=hint)
             continue
+        branches = (path.branch, opts.get(b'branch', []))
         revs, checkout = hg.addbranchrevs(
             repo, other, branches, opts.get(b"rev")
         )
@@ -4085,29 +3741,29 b' def debugbackupbundle(ui, repo, *pats, *'
                 with repo.lock(), repo.transaction(b"unbundle") as tr:
                     if scmutil.isrevsymbol(other, recovernode):
                         ui.status(_(b"Unbundling %s\n") % (recovernode))
-                        f = hg.openpath(ui, source)
-                        gen = exchange.readbundle(ui, f, source)
+                        f = hg.openpath(ui, path.loc)
+                        gen = exchange.readbundle(ui, f, path.loc)
                         if isinstance(gen, bundle2.unbundle20):
                             bundle2.applybundle(
                                 repo,
                                 gen,
                                 tr,
                                 source=b"unbundle",
-                                url=b"bundle:" + source,
+                                url=b"bundle:" + path.loc,
                             )
                         else:
-                            gen.apply(repo, b"unbundle", b"bundle:" + source)
+                            gen.apply(repo, b"unbundle", b"bundle:" + path.loc)
                         break
             else:
                 backupdate = encoding.strtolocal(
                     time.strftime(
                         "%a %H:%M, %Y-%m-%d",
-                        time.localtime(os.path.getmtime(source)),
+                        time.localtime(os.path.getmtime(path.loc)),
                     )
                 )
                 ui.status(b"\n%s\n" % (backupdate.ljust(50)))
                 if ui.verbose:
-                    ui.status(b"%s%s\n" % (b"bundle:".ljust(13), source))
+                    ui.status(b"%s%s\n" % (b"bundle:".ljust(13), path.loc))
                 else:
                     opts[
                         b"template"
@@ -4134,8 +3790,21 b' def debugsub(ui, repo, rev=None):'
         ui.writenoi18n(b' revision %s\n' % v[1])
 
 
-@command(b'debugshell', optionalrepo=True)
-def debugshell(ui, repo):
+@command(
+    b'debugshell',
+    [
+        (
+            b'c',
+            b'command',
+            b'',
+            _(b'program passed in as a string'),
+            _(b'COMMAND'),
+        )
+    ],
+    _(b'[-c COMMAND]'),
+    optionalrepo=True,
+)
+def debugshell(ui, repo, **opts):
     """run an interactive Python interpreter
 
     The local namespace is provided with a reference to the ui and
@@ -4148,10 +3817,58 b' def debugshell(ui, repo):'
         'repo': repo,
     }
 
+    # py2exe disables initialization of the site module, which is responsible
+    # for arranging for ``quit()`` to exit the interpreter.  Manually initialize
+    # the stuff that site normally does here, so that the interpreter can be
+    # quit in a consistent manner, whether run with pyoxidizer, exewrapper.c,
+    # py.exe, or py2exe.
+    if getattr(sys, "frozen", None) == 'console_exe':
+        try:
+            import site
+
+            site.setcopyright()
+            site.sethelper()
+            site.setquit()
+        except ImportError:
+            site = None  # Keep PyCharm happy
+
+    command = opts.get('command')
+    if command:
+        compiled = code.compile_command(encoding.strfromlocal(command))
+        code.InteractiveInterpreter(locals=imported_objects).runcode(compiled)
+        return
+
     code.interact(local=imported_objects)
 
 
 @command(
+    b'debug-revlog-stats',
+    [
+        (b'c', b'changelog', None, _(b'Display changelog statistics')),
+        (b'm', b'manifest', None, _(b'Display manifest statistics')),
+        (b'f', b'filelogs', None, _(b'Display filelogs statistics')),
+    ]
+    + cmdutil.formatteropts,
+)
+def debug_revlog_stats(ui, repo, **opts):
+    """display statistics about revlogs in the store"""
+    opts = pycompat.byteskwargs(opts)
+    changelog = opts[b"changelog"]
+    manifest = opts[b"manifest"]
+    filelogs = opts[b"filelogs"]
+
+    if changelog is None and manifest is None and filelogs is None:
+        changelog = True
+        manifest = True
+        filelogs = True
+
+    repo = repo.unfiltered()
+    fm = ui.formatter(b'debug-revlog-stats', opts)
+    revlog_debug.debug_revlog_stats(repo, fm, changelog, manifest, filelogs)
+    fm.end()
+
+
+@command(
     b'debugsuccessorssets',
     [(b'', b'closest', False, _(b'return closest successors sets only'))],
     _(b'[REV]'),
@@ -4843,7 +4560,8 b' def debugwireproto(ui, repo, path=None, '
                 _(b'--peer %s not supported with HTTP peers') % opts[b'peer']
             )
         else:
-            peer = httppeer.makepeer(ui, path, opener=opener)
+            peer_path = urlutil.try_path(ui, path)
+            peer = httppeer.makepeer(ui, peer_path, opener=opener)
 
         # We /could/ populate stdin/stdout with sock.makefile()...
     else:

mercurial/diffutil.py

@@ -120,7 +120,7 b' def difffeatureopts('
         )
         buildopts[b'ignorewseol'] = get(b'ignore_space_at_eol', b'ignorewseol')
     if formatchanging:
-        buildopts[b'text'] = opts and opts.get(b'text')
+        buildopts[b'text'] = None if opts is None else opts.get(b'text')
         binary = None if opts is None else opts.get(b'binary')
         buildopts[b'nobinary'] = (
             not binary

mercurial/dirstate.py

@@ -31,7 +31,6 b' from . import ('
 )
 
 from .dirstateutils import (
-    docket as docketmod,
     timestamp,
 )
 
@@ -66,10 +65,17 b' class rootcache(filecache):'
         return obj._join(fname)
 
 
-def requires_parents_change(func):
+def check_invalidated(func):
+    """check that the func is called with a non-invalidated dirstate
+
+    The dirstate is in an "invalidated state" after an error occured during its
+    modification and remains so until we exited the top level scope that framed
+    such change.
+    """
+
     def wrap(self, *args, **kwargs):
-        if not self.pendingparentchange():
-            msg = 'calling `%s` outside of a parentchange context'
+        if self._invalidated_context:
+            msg = 'calling `%s` after the dirstate was invalidated'
             msg %= func.__name__
             raise error.ProgrammingError(msg)
         return func(self, *args, **kwargs)
@@ -77,19 +83,63 b' def requires_parents_change(func):'
     return wrap
 
 
-def requires_no_parents_change(func):
+def requires_changing_parents(func):
     def wrap(self, *args, **kwargs):
-        if self.pendingparentchange():
-            msg = 'calling `%s` inside of a parentchange context'
+        if not self.is_changing_parents:
+            msg = 'calling `%s` outside of a changing_parents context'
+            msg %= func.__name__
+            raise error.ProgrammingError(msg)
+        return func(self, *args, **kwargs)
+
+    return check_invalidated(wrap)
+
+
+def requires_changing_files(func):
+    def wrap(self, *args, **kwargs):
+        if not self.is_changing_files:
+            msg = 'calling `%s` outside of a `changing_files`'
             msg %= func.__name__
             raise error.ProgrammingError(msg)
         return func(self, *args, **kwargs)
 
-    return wrap
+    return check_invalidated(wrap)
+
+
+def requires_changing_any(func):
+    def wrap(self, *args, **kwargs):
+        if not self.is_changing_any:
+            msg = 'calling `%s` outside of a changing context'
+            msg %= func.__name__
+            raise error.ProgrammingError(msg)
+        return func(self, *args, **kwargs)
+
+    return check_invalidated(wrap)
+
+
+def requires_changing_files_or_status(func):
+    def wrap(self, *args, **kwargs):
+        if not (self.is_changing_files or self._running_status > 0):
+            msg = (
+                'calling `%s` outside of a changing_files '
+                'or running_status context'
+            )
+            msg %= func.__name__
+            raise error.ProgrammingError(msg)
+        return func(self, *args, **kwargs)
+
+    return check_invalidated(wrap)
+
+
+CHANGE_TYPE_PARENTS = "parents"
+CHANGE_TYPE_FILES = "files"
 
 
 @interfaceutil.implementer(intdirstate.idirstate)
 class dirstate:
+
+    # used by largefile to avoid overwritting transaction callback
+    _tr_key_suffix = b''
+
     def __init__(
         self,
         opener,
@@ -124,7 +174,16 b' class dirstate:'
         self._dirty_tracked_set = False
         self._ui = ui
         self._filecache = {}
-        self._parentwriters = 0
+        # nesting level of `changing_parents` context
+        self._changing_level = 0
+        # the change currently underway
+        self._change_type = None
+        # number of open _running_status context
+        self._running_status = 0
+        # True if the current dirstate changing operations have been
+        # invalidated (used to make sure all nested contexts have been exited)
+        self._invalidated_context = False
+        self._attached_to_a_transaction = False
         self._filename = b'dirstate'
         self._filename_th = b'dirstate-tracked-hint'
         self._pendingfilename = b'%s.pending' % self._filename
@@ -136,6 +195,12 b' class dirstate:'
         # raises an exception).
         self._cwd
 
+    def refresh(self):
+        if '_branch' in vars(self):
+            del self._branch
+        if '_map' in vars(self) and self._map.may_need_refresh():
+            self.invalidate()
+
     def prefetch_parents(self):
         """make sure the parents are loaded
 
@@ -144,39 +209,193 b' class dirstate:'
         self._pl
 
     @contextlib.contextmanager
-    def parentchange(self):
-        """Context manager for handling dirstate parents.
+    @check_invalidated
+    def running_status(self, repo):
+        """Wrap a status operation
+
+        This context is not mutally exclusive with the `changing_*` context. It
+        also do not warrant for the `wlock` to be taken.
+
+        If the wlock is taken, this context will behave in a simple way, and
+        ensure the data are scheduled for write when leaving the top level
+        context.
 
-        If an exception occurs in the scope of the context manager,
-        the incoherent dirstate won't be written when wlock is
-        released.
+        If the lock is not taken, it will only warrant that the data are either
+        committed (written) and rolled back (invalidated) when exiting the top
+        level context. The write/invalidate action must be performed by the
+        wrapped code.
+
+
+        The expected  logic is:
+
+        A: read the dirstate
+        B: run status
+           This might make the dirstate dirty by updating cache,
+           especially in Rust.
+        C: do more "post status fixup if relevant
+        D: try to take the w-lock (this will invalidate the changes if they were raced)
+        E0: if dirstate changed on disk → discard change (done by dirstate internal)
+        E1: elif lock was acquired → write the changes
+        E2: else → discard the changes
         """
-        self._parentwriters += 1
-        yield
-        # Typically we want the "undo" step of a context manager in a
-        # finally block so it happens even when an exception
-        # occurs. In this case, however, we only want to decrement
-        # parentwriters if the code in the with statement exits
-        # normally, so we don't have a try/finally here on purpose.
-        self._parentwriters -= 1
+        has_lock = repo.currentwlock() is not None
+        is_changing = self.is_changing_any
+        tr = repo.currenttransaction()
+        has_tr = tr is not None
+        nested = bool(self._running_status)
+
+        first_and_alone = not (is_changing or has_tr or nested)
+
+        # enforce no change happened outside of a proper context.
+        if first_and_alone and self._dirty:
+            has_tr = repo.currenttransaction() is not None
+            if not has_tr and self._changing_level == 0 and self._dirty:
+                msg = "entering a status context, but dirstate is already dirty"
+                raise error.ProgrammingError(msg)
+
+        should_write = has_lock and not (nested or is_changing)
+
+        self._running_status += 1
+        try:
+            yield
+        except Exception:
+            self.invalidate()
+            raise
+        finally:
+            self._running_status -= 1
+            if self._invalidated_context:
+                should_write = False
+                self.invalidate()
+
+        if should_write:
+            assert repo.currenttransaction() is tr
+            self.write(tr)
+        elif not has_lock:
+            if self._dirty:
+                msg = b'dirstate dirty while exiting an isolated status context'
+                repo.ui.develwarn(msg)
+                self.invalidate()
+
+    @contextlib.contextmanager
+    @check_invalidated
+    def _changing(self, repo, change_type):
+        if repo.currentwlock() is None:
+            msg = b"trying to change the dirstate without holding the wlock"
+            raise error.ProgrammingError(msg)
+
+        has_tr = repo.currenttransaction() is not None
+        if not has_tr and self._changing_level == 0 and self._dirty:
+            msg = b"entering a changing context, but dirstate is already dirty"
+            repo.ui.develwarn(msg)
+
+        assert self._changing_level >= 0
+        # different type of change are mutually exclusive
+        if self._change_type is None:
+            assert self._changing_level == 0
+            self._change_type = change_type
+        elif self._change_type != change_type:
+            msg = (
+                'trying to open "%s" dirstate-changing context while a "%s" is'
+                ' already open'
+            )
+            msg %= (change_type, self._change_type)
+            raise error.ProgrammingError(msg)
+        should_write = False
+        self._changing_level += 1
+        try:
+            yield
+        except:  # re-raises
+            self.invalidate()  # this will set `_invalidated_context`
+            raise
+        finally:
+            assert self._changing_level > 0
+            self._changing_level -= 1
+            # If the dirstate is being invalidated, call invalidate again.
+            # This will throw away anything added by a upper context and
+            # reset the `_invalidated_context` flag when relevant
+            if self._changing_level <= 0:
+                self._change_type = None
+                assert self._changing_level == 0
+            if self._invalidated_context:
+                # make sure we invalidate anything an upper context might
+                # have changed.
+                self.invalidate()
+            else:
+                should_write = self._changing_level <= 0
+        tr = repo.currenttransaction()
+        if has_tr != (tr is not None):
+            if has_tr:
+                m = "transaction vanished while changing dirstate"
+            else:
+                m = "transaction appeared while changing dirstate"
+            raise error.ProgrammingError(m)
+        if should_write:
+            self.write(tr)
+
+    @contextlib.contextmanager
+    def changing_parents(self, repo):
+        with self._changing(repo, CHANGE_TYPE_PARENTS) as c:
+            yield c
+
+    @contextlib.contextmanager
+    def changing_files(self, repo):
+        with self._changing(repo, CHANGE_TYPE_FILES) as c:
+            yield c
+
+    # here to help migration to the new code
+    def parentchange(self):
+        msg = (
+            "Mercurial 6.4 and later requires call to "
+            "`dirstate.changing_parents(repo)`"
+        )
+        raise error.ProgrammingError(msg)
+
+    @property
+    def is_changing_any(self):
+        """Returns true if the dirstate is in the middle of a set of changes.
+
+        This returns True for any kind of change.
+        """
+        return self._changing_level > 0
 
     def pendingparentchange(self):
+        return self.is_changing_parent()
+
+    def is_changing_parent(self):
         """Returns true if the dirstate is in the middle of a set of changes
         that modify the dirstate parent.
         """
-        return self._parentwriters > 0
+        self._ui.deprecwarn(b"dirstate.is_changing_parents", b"6.5")
+        return self.is_changing_parents
+
+    @property
+    def is_changing_parents(self):
+        """Returns true if the dirstate is in the middle of a set of changes
+        that modify the dirstate parent.
+        """
+        if self._changing_level <= 0:
+            return False
+        return self._change_type == CHANGE_TYPE_PARENTS
+
+    @property
+    def is_changing_files(self):
+        """Returns true if the dirstate is in the middle of a set of changes
+        that modify the files tracked or their sources.
+        """
+        if self._changing_level <= 0:
+            return False
+        return self._change_type == CHANGE_TYPE_FILES
 
     @propertycache
     def _map(self):
         """Return the dirstate contents (see documentation for dirstatemap)."""
-        self._map = self._mapcls(
+        return self._mapcls(
             self._ui,
             self._opener,
             self._root,
             self._nodeconstants,
             self._use_dirstate_v2,
         )
-        return self._map
 
     @property
     def _sparsematcher(self):
@@ -365,6 +584,7 b' class dirstate:'
     def branch(self):
         return encoding.tolocal(self._branch)
 
+    @requires_changing_parents
     def setparents(self, p1, p2=None):
         """Set dirstate parents to p1 and p2.
 
@@ -376,10 +596,10 b' class dirstate:'
         """
         if p2 is None:
             p2 = self._nodeconstants.nullid
-        if self._parentwriters == 0:
+        if self._changing_level == 0:
             raise ValueError(
                 b"cannot set dirstate parent outside of "
-                b"dirstate.parentchange context manager"
+                b"dirstate.changing_parents context manager"
             )
 
         self._dirty = True
@@ -419,9 +639,14 b' class dirstate:'
                 delattr(self, a)
         self._dirty = False
         self._dirty_tracked_set = False
-        self._parentwriters = 0
+        self._invalidated_context = bool(
+            self._changing_level > 0
+            or self._attached_to_a_transaction
+            or self._running_status
+        )
         self._origpl = None
 
+    @requires_changing_any
     def copy(self, source, dest):
         """Mark dest as a copy of source. Unmark dest if source is None."""
         if source == dest:
@@ -439,7 +664,7 b' class dirstate:'
     def copies(self):
         return self._map.copymap
 
-    @requires_no_parents_change
+    @requires_changing_files
     def set_tracked(self, filename, reset_copy=False):
         """a "public" method for generic code to mark a file as tracked
 
@@ -461,7 +686,7 b' class dirstate:'
             self._dirty_tracked_set = True
         return pre_tracked
 
-    @requires_no_parents_change
+    @requires_changing_files
     def set_untracked(self, filename):
         """a "public" method for generic code to mark a file as untracked
 
@@ -476,7 +701,7 b' class dirstate:'
             self._dirty_tracked_set = True
         return ret
 
-    @requires_no_parents_change
+    @requires_changing_files_or_status
     def set_clean(self, filename, parentfiledata):
         """record that the current state of the file on disk is known to be clean"""
         self._dirty = True
@@ -485,13 +710,13 b' class dirstate:'
         (mode, size, mtime) = parentfiledata
         self._map.set_clean(filename, mode, size, mtime)
 
-    @requires_no_parents_change
+    @requires_changing_files_or_status
     def set_possibly_dirty(self, filename):
         """record that the current state of the file on disk is unknown"""
         self._dirty = True
         self._map.set_possibly_dirty(filename)
 
-    @requires_parents_change
+    @requires_changing_parents
     def update_file_p1(
         self,
         filename,
@@ -503,7 +728,7 b' class dirstate:'
         rewriting operation.
 
         It should not be called during a merge (p2 != nullid) and only within
-        a `with dirstate.parentchange():` context.
+        a `with dirstate.changing_parents(repo):` context.
         """
         if self.in_merge:
             msg = b'update_file_reference should not be called when merging'
@@ -531,7 +756,7 b' class dirstate:'
             has_meaningful_mtime=False,
         )
 
-    @requires_parents_change
+    @requires_changing_parents
     def update_file(
         self,
         filename,
@@ -546,12 +771,57 b' class dirstate:'
         This is to be called when the direstates parent changes to keep track
         of what is the file situation in regards to the working copy and its parent.
 
-        This function must be called within a `dirstate.parentchange` context.
+        This function must be called within a `dirstate.changing_parents` context.
 
         note: the API is at an early stage and we might need to adjust it
         depending of what information ends up being relevant and useful to
         other processing.
         """
+        self._update_file(
+            filename=filename,
+            wc_tracked=wc_tracked,
+            p1_tracked=p1_tracked,
+            p2_info=p2_info,
+            possibly_dirty=possibly_dirty,
+            parentfiledata=parentfiledata,
+        )
+
+    def hacky_extension_update_file(self, *args, **kwargs):
+        """NEVER USE THIS, YOU DO NOT NEED IT
+
+        This function is a variant of "update_file" to be called by a small set
+        of extensions, it also adjust the internal state of file, but can be
+        called outside an `changing_parents` context.
+
+        A very small number of extension meddle with the working copy content
+        in a way that requires to adjust the dirstate accordingly. At the time
+        this command is written they are :
+        - keyword,
+        - largefile,
+        PLEASE DO NOT GROW THIS LIST ANY FURTHER.
+
+        This function could probably be replaced by more semantic one (like
+        "adjust expected size" or "always revalidate file content", etc)
+        however at the time where this is writen, this is too much of a detour
+        to be considered.
+        """
+        if not (self._changing_level > 0 or self._running_status > 0):
+            msg = "requires a changes context"
+            raise error.ProgrammingError(msg)
+        self._update_file(
+            *args,
+            **kwargs,
+        )
+
+    def _update_file(
+        self,
+        filename,
+        wc_tracked,
+        p1_tracked,
+        p2_info=False,
+        possibly_dirty=False,
+        parentfiledata=None,
+    ):
 
         # note: I do not think we need to double check name clash here since we
         # are in a update/merge case that should already have taken care of
@@ -680,12 +950,16 b' class dirstate:'
             return self._normalize(path, isknown, ignoremissing)
         return path
 
+    # XXX this method is barely used, as a result:
+    # - its semantic is unclear
+    # - do we really needs it ?
+    @requires_changing_parents
     def clear(self):
         self._map.clear()
         self._dirty = True
 
+    @requires_changing_parents
     def rebuild(self, parent, allfiles, changedfiles=None):
-
         matcher = self._sparsematcher
         if matcher is not None and not matcher.always():
             # should not add non-matching files
@@ -724,7 +998,6 b' class dirstate:'
         self._map.setparents(parent, self._nodeconstants.nullid)
 
         for f in to_lookup:
-
             if self.in_merge:
                 self.set_tracked(f)
             else:
@@ -749,20 +1022,41 b' class dirstate:'
     def write(self, tr):
         if not self._dirty:
             return
+        # make sure we don't request a write of invalidated content
+        # XXX move before the dirty check once `unlock` stop calling `write`
+        assert not self._invalidated_context
 
         write_key = self._use_tracked_hint and self._dirty_tracked_set
         if tr:
+
+            def on_abort(tr):
+                self._attached_to_a_transaction = False
+                self.invalidate()
+
+            # make sure we invalidate the current change on abort
+            if tr is not None:
+                tr.addabort(
+                    b'dirstate-invalidate%s' % self._tr_key_suffix,
+                    on_abort,
+                )
+
+            self._attached_to_a_transaction = True
+
+            def on_success(f):
+                self._attached_to_a_transaction = False
+                self._writedirstate(tr, f),
+
             # delay writing in-memory changes out
             tr.addfilegenerator(
-                b'dirstate-1-main',
+                b'dirstate-1-main%s' % self._tr_key_suffix,
                 (self._filename,),
-                lambda f: self._writedirstate(tr, f),
+                on_success,
                 location=b'plain',
                 post_finalize=True,
             )
             if write_key:
                 tr.addfilegenerator(
-                    b'dirstate-2-key-post',
+                    b'dirstate-2-key-post%s' % self._tr_key_suffix,
                     (self._filename_th,),
                     lambda f: self._write_tracked_hint(tr, f),
                     location=b'plain',
@@ -798,6 +1092,8 b' class dirstate:'
         self._plchangecallbacks[category] = callback
 
     def _writedirstate(self, tr, st):
+        # make sure we don't write invalidated content
+        assert not self._invalidated_context
         # notify callbacks about parents change
         if self._origpl is not None and self._origpl != self._pl:
             for c, callback in sorted(self._plchangecallbacks.items()):
@@ -936,7 +1232,8 b' class dirstate:'
                     badfn(ff, badtype(kind))
                     if nf in dmap:
                         results[nf] = None
-            except OSError as inst:  # nf not found on disk - it is dirstate only
+            except (OSError) as inst:
+                # nf not found on disk - it is dirstate only
                 if nf in dmap:  # does it exactly match a missing file?
                     results[nf] = None
                 else:  # does it match a missing directory?
@@ -1246,7 +1543,7 b' class dirstate:'
                         )
                     )
 
-        for (fn, message) in bad:
+        for fn, message in bad:
             matcher.bad(fn, encoding.strtolocal(message))
 
         status = scmutil.status(
@@ -1276,6 +1573,9 b' class dirstate:'
             files that have definitely not been modified since the
             dirstate was written
         """
+        if not self._running_status:
+            msg = "Calling `status` outside a `running_status` context"
+            raise error.ProgrammingError(msg)
         listignored, listclean, listunknown = ignored, clean, unknown
         lookup, modified, added, unknown, ignored = [], [], [], [], []
         removed, deleted, clean = [], [], []
@@ -1435,142 +1735,47 b' class dirstate:'
         else:
             return self._filename
 
-    def data_backup_filename(self, backupname):
-        if not self._use_dirstate_v2:
-            return None
-        return backupname + b'.v2-data'
-
-    def _new_backup_data_filename(self, backupname):
-        """return a filename to backup a data-file or None"""
-        if not self._use_dirstate_v2:
-            return None
-        if self._map.docket.uuid is None:
-            # not created yet, nothing to backup
-            return None
-        data_filename = self._map.docket.data_filename()
-        return data_filename, self.data_backup_filename(backupname)
-
-    def backup_data_file(self, backupname):
-        if not self._use_dirstate_v2:
-            return None
-        docket = docketmod.DirstateDocket.parse(
-            self._opener.read(backupname),
-            self._nodeconstants,
-        )
-        return self.data_backup_filename(backupname), docket.data_filename()
-
-    def savebackup(self, tr, backupname):
-        '''Save current dirstate into backup file'''
-        filename = self._actualfilename(tr)
-        assert backupname != filename
+    def all_file_names(self):
+        """list all filename currently used by this dirstate
 
-        # use '_writedirstate' instead of 'write' to write changes certainly,
-        # because the latter omits writing out if transaction is running.
-        # output file will be used to create backup of dirstate at this point.
-        if self._dirty or not self._opener.exists(filename):
-            self._writedirstate(
-                tr,
-                self._opener(filename, b"w", atomictemp=True, checkambig=True),
+        This is only used to do `hg rollback` related backup in the transaction
+        """
+        if not self._opener.exists(self._filename):
+            # no data every written to disk yet
+            return ()
+        elif self._use_dirstate_v2:
+            return (
+                self._filename,
+                self._map.docket.data_filename(),
             )
+        else:
+            return (self._filename,)
 
-        if tr:
-            # ensure that subsequent tr.writepending returns True for
-            # changes written out above, even if dirstate is never
-            # changed after this
-            tr.addfilegenerator(
-                b'dirstate-1-main',
-                (self._filename,),
-                lambda f: self._writedirstate(tr, f),
-                location=b'plain',
-                post_finalize=True,
-            )
-
-            # ensure that pending file written above is unlinked at
-            # failure, even if tr.writepending isn't invoked until the
-            # end of this transaction
-            tr.registertmp(filename, location=b'plain')
-
-        self._opener.tryunlink(backupname)
-        # hardlink backup is okay because _writedirstate is always called
-        # with an "atomictemp=True" file.
-        util.copyfile(
-            self._opener.join(filename),
-            self._opener.join(backupname),
-            hardlink=True,
+    def verify(self, m1, m2, p1, narrow_matcher=None):
+        """
+        check the dirstate contents against the parent manifest and yield errors
+        """
+        missing_from_p1 = _(
+            b"%s marked as tracked in p1 (%s) but not in manifest1\n"
         )
-        data_pair = self._new_backup_data_filename(backupname)
-        if data_pair is not None:
-            data_filename, bck_data_filename = data_pair
-            util.copyfile(
-                self._opener.join(data_filename),
-                self._opener.join(bck_data_filename),
-                hardlink=True,
-            )
-            if tr is not None:
-                # ensure that pending file written above is unlinked at
-                # failure, even if tr.writepending isn't invoked until the
-                # end of this transaction
-                tr.registertmp(bck_data_filename, location=b'plain')
-
-    def restorebackup(self, tr, backupname):
-        '''Restore dirstate by backup file'''
-        # this "invalidate()" prevents "wlock.release()" from writing
-        # changes of dirstate out after restoring from backup file
-        self.invalidate()
-        o = self._opener
-        if not o.exists(backupname):
-            # there was no file backup, delete existing files
-            filename = self._actualfilename(tr)
-            data_file = None
-            if self._use_dirstate_v2 and self._map.docket.uuid is not None:
-                data_file = self._map.docket.data_filename()
-            if o.exists(filename):
-                o.unlink(filename)
-            if data_file is not None and o.exists(data_file):
-                o.unlink(data_file)
-            return
-        filename = self._actualfilename(tr)
-        data_pair = self.backup_data_file(backupname)
-        if o.exists(filename) and util.samefile(
-            o.join(backupname), o.join(filename)
-        ):
-            o.unlink(backupname)
-        else:
-            o.rename(backupname, filename, checkambig=True)
-
-        if data_pair is not None:
-            data_backup, target = data_pair
-            if o.exists(target) and util.samefile(
-                o.join(data_backup), o.join(target)
-            ):
-                o.unlink(data_backup)
-            else:
-                o.rename(data_backup, target, checkambig=True)
-
-    def clearbackup(self, tr, backupname):
-        '''Clear backup file'''
-        o = self._opener
-        if o.exists(backupname):
-            data_backup = self.backup_data_file(backupname)
-            o.unlink(backupname)
-            if data_backup is not None:
-                o.unlink(data_backup[0])
-
-    def verify(self, m1, m2):
-        """check the dirstate content again the parent manifest and yield errors"""
-        missing_from_p1 = b"%s in state %s, but not in manifest1\n"
-        unexpected_in_p1 = b"%s in state %s, but also in manifest1\n"
-        missing_from_ps = b"%s in state %s, but not in either manifest\n"
-        missing_from_ds = b"%s in manifest1, but listed as state %s\n"
+        unexpected_in_p1 = _(b"%s marked as added, but also in manifest1\n")
+        missing_from_ps = _(
+            b"%s marked as modified, but not in either manifest\n"
+        )
+        missing_from_ds = _(
+            b"%s in manifest1, but not marked as tracked in p1 (%s)\n"
+        )
         for f, entry in self.items():
-            state = entry.state
-            if state in b"nr" and f not in m1:
-                yield (missing_from_p1, f, state)
-            if state in b"a" and f in m1:
-                yield (unexpected_in_p1, f, state)
-            if state in b"m" and f not in m1 and f not in m2:
-                yield (missing_from_ps, f, state)
+            if entry.p1_tracked:
+                if entry.modified and f not in m1 and f not in m2:
+                    yield missing_from_ps % f
+                elif f not in m1:
+                    yield missing_from_p1 % (f, node.short(p1))
+            if entry.added and f in m1:
+                yield unexpected_in_p1 % f
         for f in m1:
-            state = self.get_entry(f).state
-            if state not in b"nrm":
-                yield (missing_from_ds, f, state)
+            if narrow_matcher is not None and not narrow_matcher(f):
+                continue
+            entry = self.get_entry(f)
+            if not entry.p1_tracked:
+                yield missing_from_ds % (f, node.short(p1))

mercurial/dirstatemap.py

@@ -58,6 +58,34 b' class _dirstatemapcommon:'
         # for consistent view between _pl() and _read() invocations
         self._pendingmode = None
 
+    def _set_identity(self):
+        self.identity = self._get_current_identity()
+
+    def _get_current_identity(self):
+        try:
+            return util.cachestat(self._opener.join(self._filename))
+        except FileNotFoundError:
+            return None
+
+    def may_need_refresh(self):
+        if 'identity' not in vars(self):
+            # no existing identity, we need a refresh
+            return True
+        if self.identity is None:
+            return True
+        if not self.identity.cacheable():
+            # We cannot trust the entry
+            # XXX this is a problem on windows, NFS, or other inode less system
+            return True
+        current_identity = self._get_current_identity()
+        if current_identity is None:
+            return True
+        if not current_identity.cacheable():
+            # We cannot trust the entry
+            # XXX this is a problem on windows, NFS, or other inode less system
+            return True
+        return current_identity != self.identity
+
     def preload(self):
         """Loads the underlying data, if it's not already loaded"""
         self._map
@@ -118,6 +146,9 b' class _dirstatemapcommon:'
             raise error.ProgrammingError(b'dirstate docket name collision')
         data_filename = new_docket.data_filename()
         self._opener.write(data_filename, packed)
+        # tell the transaction that we are adding a new file
+        if tr is not None:
+            tr.addbackup(data_filename, location=b'plain')
         # Write the new docket after the new data file has been
         # written. Because `st` was opened with `atomictemp=True`,
         # the actual `.hg/dirstate` file is only affected on close.
@@ -127,6 +158,8 b' class _dirstatemapcommon:'
         # the new data file was written.
         if old_docket.uuid:
             data_filename = old_docket.data_filename()
+            if tr is not None:
+                tr.addbackup(data_filename, location=b'plain')
             unlink = lambda _tr=None: self._opener.unlink(data_filename)
             if tr:
                 category = b"dirstate-v2-clean-" + old_docket.uuid
@@ -258,9 +291,7 b' class dirstatemap(_dirstatemapcommon):'
 
     def read(self):
         # ignore HG_PENDING because identity is used only for writing
-        self.identity = util.filestat.frompath(
-            self._opener.join(self._filename)
-        )
+        self._set_identity()
 
         if self._use_dirstate_v2:
             if not self.docket.uuid:
@@ -523,9 +554,7 b' if rustmod is not None:'
             Fills the Dirstatemap when called.
             """
             # ignore HG_PENDING because identity is used only for writing
-            self.identity = util.filestat.frompath(
-                self._opener.join(self._filename)
-            )
+            self._set_identity()
 
             if self._use_dirstate_v2:
                 if self.docket.uuid:
@@ -614,6 +643,14 b' if rustmod is not None:'
             if append:
                 docket = self.docket
                 data_filename = docket.data_filename()
+                # We mark it for backup to make sure a future `hg rollback` (or
+                # `hg recover`?) call find the data it needs to restore a
+                # working repository.
+                #
+                # The backup can use a hardlink because the format is resistant
+                # to trailing "dead" data.
+                if tr is not None:
+                    tr.addbackup(data_filename, location=b'plain')
                 with self._opener(data_filename, b'r+b') as fp:
                     fp.seek(docket.data_size)
                     assert fp.tell() == docket.data_size

mercurial/dispatch.py

@@ -980,7 +980,8 b' def _getlocal(ui, rpath, wd=None):'
             lui.readconfig(os.path.join(path, b".hg", b"hgrc-not-shared"), path)
 
     if rpath:
-        path = urlutil.get_clone_path(lui, rpath)[0]
+        path_obj = urlutil.get_clone_path_obj(lui, rpath)
+        path = path_obj.rawloc
         lui = ui.copy()
         if rcutil.use_repo_hgrc():
             _readsharedsourceconfig(lui, path)

mercurial/exchange.py

@@ -1183,7 +1183,12 b' def _pushbundle2(pushop):'
             trgetter = None
             if pushback:
                 trgetter = pushop.trmanager.transaction
-            op = bundle2.processbundle(pushop.repo, reply, trgetter)
+            op = bundle2.processbundle(
+                pushop.repo,
+                reply,
+                trgetter,
+                remote=pushop.remote,
+            )
         except error.BundleValueError as exc:
             raise error.RemoteError(_(b'missing support for %s') % exc)
         except bundle2.AbortFromPart as exc:
@@ -1903,10 +1908,18 b' def _pullbundle2(pullop):'
 
         try:
             op = bundle2.bundleoperation(
-                pullop.repo, pullop.gettransaction, source=b'pull'
+                pullop.repo,
+                pullop.gettransaction,
+                source=b'pull',
+                remote=pullop.remote,
             )
             op.modes[b'bookmarks'] = b'records'
-            bundle2.processbundle(pullop.repo, bundle, op=op)
+            bundle2.processbundle(
+                pullop.repo,
+                bundle,
+                op=op,
+                remote=pullop.remote,
+            )
         except bundle2.AbortFromPart as exc:
             pullop.repo.ui.error(_(b'remote: abort: %s\n') % exc)
             raise error.RemoteError(_(b'pull failed on remote'), hint=exc.hint)
@@ -1995,7 +2008,12 b' def _pullchangeset(pullop):'
             ).result()
 
     bundleop = bundle2.applybundle(
-        pullop.repo, cg, tr, b'pull', pullop.remote.url()
+        pullop.repo,
+        cg,
+        tr,
+        b'pull',
+        pullop.remote.url(),
+        remote=pullop.remote,
     )
     pullop.cgresult = bundle2.combinechangegroupresults(bundleop)
 

mercurial/filelog.py

@@ -111,6 +111,7 b' class filelog:'
         assumehaveparentrevisions=False,
         deltamode=repository.CG_DELTAMODE_STD,
         sidedata_helpers=None,
+        debug_info=None,
     ):
         return self._revlog.emitrevisions(
             nodes,
@@ -119,6 +120,7 b' class filelog:'
             assumehaveparentrevisions=assumehaveparentrevisions,
             deltamode=deltamode,
             sidedata_helpers=sidedata_helpers,
+            debug_info=debug_info,
         )
 
     def addrevision(
@@ -151,6 +153,8 b' class filelog:'
         addrevisioncb=None,
         duplicaterevisioncb=None,
         maybemissingparents=False,
+        debug_info=None,
+        delta_base_reuse_policy=None,
     ):
         if maybemissingparents:
             raise error.Abort(
@@ -171,6 +175,8 b' class filelog:'
                 transaction,
                 addrevisioncb=addrevisioncb,
                 duplicaterevisioncb=duplicaterevisioncb,
+                debug_info=debug_info,
+                delta_base_reuse_policy=delta_base_reuse_policy,
             )
 
     def getstrippoint(self, minlink):

mercurial/filemerge.py

@@ -158,7 +158,7 b' def findexternaltool(ui, tool):'
             continue
         p = util.lookupreg(k, _toolstr(ui, tool, b"regname"))
         if p:
-            p = procutil.findexe(p + _toolstr(ui, tool, b"regappend", b""))
+            p = procutil.findexe(p + _toolstr(ui, tool, b"regappend"))
             if p:
                 return p
     exe = _toolstr(ui, tool, b"executable", tool)
@@ -478,8 +478,9 b' def _merge(repo, local, other, base, mod'
     """
     Uses the internal non-interactive simple merge algorithm for merging
     files. It will fail if there are any conflicts and leave markers in
-    the partially merged file. Markers will have two sections, one for each side
-    of merge, unless mode equals 'union' which suppresses the markers."""
+    the partially merged file. Markers will have two sections, one for each
+    side of merge, unless mode equals 'union' or 'union-other-first' which
+    suppresses the markers."""
     ui = repo.ui
 
     try:
@@ -510,12 +511,28 b' def _merge(repo, local, other, base, mod'
 def _iunion(repo, mynode, local, other, base, toolconf, backup):
     """
     Uses the internal non-interactive simple merge algorithm for merging
-    files. It will use both left and right sides for conflict regions.
+    files. It will use both local and other sides for conflict regions by
+    adding local on top of other.
     No markers are inserted."""
     return _merge(repo, local, other, base, b'union')
 
 
 @internaltool(
+    b'union-other-first',
+    fullmerge,
+    _(
+        b"warning: conflicts while merging %s! "
+        b"(edit, then use 'hg resolve --mark')\n"
+    ),
+    precheck=_mergecheck,
+)
+def _iunion_other_first(repo, mynode, local, other, base, toolconf, backup):
+    """
+    Like :union, but add other on top of local."""
+    return _merge(repo, local, other, base, b'union-other-first')
+
+
+@internaltool(
     b'merge',
     fullmerge,
     _(

mercurial/help.py

@@ -10,6 +10,18 b' import itertools'
 import re
 import textwrap
 
+from typing import (
+    Callable,
+    Dict,
+    Iterable,
+    List,
+    Optional,
+    Set,
+    Tuple,
+    Union,
+    cast,
+)
+
 from .i18n import (
     _,
     gettext,
@@ -40,7 +52,16 b' from .utils import ('
     stringutil,
 )
 
-_exclkeywords = {
+_DocLoader = Callable[[uimod.ui], bytes]
+# Old extensions may not register with a category
+_HelpEntry = Union["_HelpEntryNoCategory", "_HelpEntryWithCategory"]
+_HelpEntryNoCategory = Tuple[List[bytes], bytes, _DocLoader]
+_HelpEntryWithCategory = Tuple[List[bytes], bytes, _DocLoader, bytes]
+_SelectFn = Callable[[object], bool]
+_SynonymTable = Dict[bytes, List[bytes]]
+_TopicHook = Callable[[uimod.ui, bytes, bytes], bytes]
+
+_exclkeywords: Set[bytes] = {
     b"(ADVANCED)",
     b"(DEPRECATED)",
     b"(EXPERIMENTAL)",
@@ -56,7 +77,7 b' from .utils import ('
 # Extensions with custom categories should insert them into this list
 # after/before the appropriate item, rather than replacing the list or
 # assuming absolute positions.
-CATEGORY_ORDER = [
+CATEGORY_ORDER: List[bytes] = [
     registrar.command.CATEGORY_REPO_CREATION,
     registrar.command.CATEGORY_REMOTE_REPO_MANAGEMENT,
     registrar.command.CATEGORY_COMMITTING,
@@ -74,7 +95,7 b' CATEGORY_ORDER = ['
 
 # Human-readable category names. These are translated.
 # Extensions with custom categories should add their names here.
-CATEGORY_NAMES = {
+CATEGORY_NAMES: Dict[bytes, bytes] = {
     registrar.command.CATEGORY_REPO_CREATION: b'Repository creation',
     registrar.command.CATEGORY_REMOTE_REPO_MANAGEMENT: b'Remote repository management',
     registrar.command.CATEGORY_COMMITTING: b'Change creation',
@@ -102,7 +123,7 b" TOPIC_CATEGORY_NONE = b'none'"
 # Extensions with custom categories should insert them into this list
 # after/before the appropriate item, rather than replacing the list or
 # assuming absolute positions.
-TOPIC_CATEGORY_ORDER = [
+TOPIC_CATEGORY_ORDER: List[bytes] = [
     TOPIC_CATEGORY_IDS,
     TOPIC_CATEGORY_OUTPUT,
     TOPIC_CATEGORY_CONFIG,
@@ -112,7 +133,7 b' TOPIC_CATEGORY_ORDER = ['
 ]
 
 # Human-readable topic category names. These are translated.
-TOPIC_CATEGORY_NAMES = {
+TOPIC_CATEGORY_NAMES: Dict[bytes, bytes] = {
     TOPIC_CATEGORY_IDS: b'Mercurial identifiers',
     TOPIC_CATEGORY_OUTPUT: b'Mercurial output',
     TOPIC_CATEGORY_CONFIG: b'Mercurial configuration',
@@ -122,7 +143,12 b' TOPIC_CATEGORY_NAMES = {'
 }
 
 
-def listexts(header, exts, indent=1, showdeprecated=False):
+def listexts(
+    header: bytes,
+    exts: Dict[bytes, bytes],
+    indent: int = 1,
+    showdeprecated: bool = False,
+) -> List[bytes]:
     '''return a text listing of the given extensions'''
     rst = []
     if exts:
@@ -135,7 +161,7 b' def listexts(header, exts, indent=1, sho'
     return rst
 
 
-def extshelp(ui):
+def extshelp(ui: uimod.ui) -> bytes:
     rst = loaddoc(b'extensions')(ui).splitlines(True)
     rst.extend(
         listexts(
@@ -153,7 +179,7 b' def extshelp(ui):'
     return doc
 
 
-def parsedefaultmarker(text):
+def parsedefaultmarker(text: bytes) -> Optional[Tuple[bytes, List[bytes]]]:
     """given a text 'abc (DEFAULT: def.ghi)',
     returns (b'abc', (b'def', b'ghi')). Otherwise return None"""
     if text[-1:] == b')':
@@ -164,7 +190,7 b' def parsedefaultmarker(text):'
             return text[:pos], item.split(b'.', 2)
 
 
-def optrst(header, options, verbose, ui):
+def optrst(header: bytes, options, verbose: bool, ui: uimod.ui) -> bytes:
     data = []
     multioccur = False
     for option in options:
@@ -220,13 +246,15 b' def optrst(header, options, verbose, ui)'
     return b''.join(rst)
 
 
-def indicateomitted(rst, omitted, notomitted=None):
+def indicateomitted(
+    rst: List[bytes], omitted: bytes, notomitted: Optional[bytes] = None
+) -> None:
     rst.append(b'\n\n.. container:: omitted\n\n    %s\n\n' % omitted)
     if notomitted:
         rst.append(b'\n\n.. container:: notomitted\n\n    %s\n\n' % notomitted)
 
 
-def filtercmd(ui, cmd, func, kw, doc):
+def filtercmd(ui: uimod.ui, cmd: bytes, func, kw: bytes, doc: bytes) -> bool:
     if not ui.debugflag and cmd.startswith(b"debug") and kw != b"debug":
         # Debug command, and user is not looking for those.
         return True
@@ -249,11 +277,13 b' def filtercmd(ui, cmd, func, kw, doc):'
     return False
 
 
-def filtertopic(ui, topic):
+def filtertopic(ui: uimod.ui, topic: bytes) -> bool:
     return ui.configbool(b'help', b'hidden-topic.%s' % topic, False)
 
 
-def topicmatch(ui, commands, kw):
+def topicmatch(
+    ui: uimod.ui, commands, kw: bytes
+) -> Dict[bytes, List[Tuple[bytes, bytes]]]:
     """Return help topics matching kw.
 
     Returns {'section': [(name, summary), ...], ...} where section is
@@ -326,10 +356,10 b' def topicmatch(ui, commands, kw):'
     return results
 
 
-def loaddoc(topic, subdir=None):
+def loaddoc(topic: bytes, subdir: Optional[bytes] = None) -> _DocLoader:
     """Return a delayed loader for help/topic.txt."""
 
-    def loader(ui):
+    def loader(ui: uimod.ui) -> bytes:
         package = b'mercurial.helptext'
         if subdir:
             package += b'.' + subdir
@@ -342,7 +372,7 b' def loaddoc(topic, subdir=None):'
     return loader
 
 
-internalstable = sorted(
+internalstable: List[_HelpEntryNoCategory] = sorted(
     [
         (
             [b'bid-merge'],
@@ -407,7 +437,7 b' internalstable = sorted('
 )
 
 
-def internalshelp(ui):
+def internalshelp(ui: uimod.ui) -> bytes:
     """Generate the index for the "internals" topic."""
     lines = [
         b'To access a subtopic, use "hg help internals.{subtopic-name}"\n',
@@ -419,7 +449,7 b' def internalshelp(ui):'
     return b''.join(lines)
 
 
-helptable = sorted(
+helptable: List[_HelpEntryWithCategory] = sorted(
     [
         (
             [b'bundlespec'],
@@ -581,20 +611,27 b' helptable = sorted('
 )
 
 # Maps topics with sub-topics to a list of their sub-topics.
-subtopics = {
+subtopics: Dict[bytes, List[_HelpEntryNoCategory]] = {
     b'internals': internalstable,
 }
 
 # Map topics to lists of callable taking the current topic help and
 # returning the updated version
-helphooks = {}
+helphooks: Dict[bytes, List[_TopicHook]] = {}
 
 
-def addtopichook(topic, rewriter):
+def addtopichook(topic: bytes, rewriter: _TopicHook) -> None:
     helphooks.setdefault(topic, []).append(rewriter)
 
 
-def makeitemsdoc(ui, topic, doc, marker, items, dedent=False):
+def makeitemsdoc(
+    ui: uimod.ui,
+    topic: bytes,
+    doc: bytes,
+    marker: bytes,
+    items: Dict[bytes, bytes],
+    dedent: bool = False,
+) -> bytes:
     """Extract docstring from the items key to function mapping, build a
     single documentation block and use it to overwrite the marker in doc.
     """
@@ -622,8 +659,10 b' def makeitemsdoc(ui, topic, doc, marker,'
     return doc.replace(marker, entries)
 
 
-def addtopicsymbols(topic, marker, symbols, dedent=False):
-    def add(ui, topic, doc):
+def addtopicsymbols(
+    topic: bytes, marker: bytes, symbols, dedent: bool = False
+) -> None:
+    def add(ui: uimod.ui, topic: bytes, doc: bytes):
         return makeitemsdoc(ui, topic, doc, marker, symbols, dedent=dedent)
 
     addtopichook(topic, add)
@@ -647,7 +686,7 b' addtopicsymbols('
 )
 
 
-def inserttweakrc(ui, topic, doc):
+def inserttweakrc(ui: uimod.ui, topic: bytes, doc: bytes) -> bytes:
     marker = b'.. tweakdefaultsmarker'
     repl = uimod.tweakrc
 
@@ -658,7 +697,9 b' def inserttweakrc(ui, topic, doc):'
     return re.sub(br'( *)%s' % re.escape(marker), sub, doc)
 
 
-def _getcategorizedhelpcmds(ui, cmdtable, name, select=None):
+def _getcategorizedhelpcmds(
+    ui: uimod.ui, cmdtable, name: bytes, select: Optional[_SelectFn] = None
+) -> Tuple[Dict[bytes, List[bytes]], Dict[bytes, bytes], _SynonymTable]:
     # Category -> list of commands
     cats = {}
     # Command -> short description
@@ -687,16 +728,18 b' def _getcategorizedhelpcmds(ui, cmdtable'
     return cats, h, syns
 
 
-def _getcategorizedhelptopics(ui, topictable):
+def _getcategorizedhelptopics(
+    ui: uimod.ui, topictable: List[_HelpEntry]
+) -> Tuple[Dict[bytes, List[Tuple[bytes, bytes]]], Dict[bytes, List[bytes]]]:
     # Group commands by category.
     topiccats = {}
     syns = {}
     for topic in topictable:
         names, header, doc = topic[0:3]
         if len(topic) > 3 and topic[3]:
-            category = topic[3]
+            category: bytes = cast(bytes, topic[3])  # help pytype
         else:
-            category = TOPIC_CATEGORY_NONE
+            category: bytes = TOPIC_CATEGORY_NONE
 
         topicname = names[0]
         syns[topicname] = list(names)
@@ -709,15 +752,15 b" addtopichook(b'config', inserttweakrc)"
 
 
 def help_(
-    ui,
+    ui: uimod.ui,
     commands,
-    name,
-    unknowncmd=False,
-    full=True,
-    subtopic=None,
-    fullname=None,
+    name: bytes,
+    unknowncmd: bool = False,
+    full: bool = True,
+    subtopic: Optional[bytes] = None,
+    fullname: Optional[bytes] = None,
     **opts
-):
+) -> bytes:
     """
     Generate the help for 'name' as unformatted restructured text. If
     'name' is None, describe the commands available.
@@ -725,7 +768,7 b' def help_('
 
     opts = pycompat.byteskwargs(opts)
 
-    def helpcmd(name, subtopic=None):
+    def helpcmd(name: bytes, subtopic: Optional[bytes]) -> List[bytes]:
         try:
             aliases, entry = cmdutil.findcmd(
                 name, commands.table, strict=unknowncmd
@@ -826,7 +869,7 b' def help_('
 
         return rst
 
-    def helplist(select=None, **opts):
+    def helplist(select: Optional[_SelectFn] = None, **opts) -> List[bytes]:
         cats, h, syns = _getcategorizedhelpcmds(
             ui, commands.table, name, select
         )
@@ -846,7 +889,7 b' def help_('
             else:
                 rst.append(_(b'list of commands:\n'))
 
-        def appendcmds(cmds):
+        def appendcmds(cmds: Iterable[bytes]) -> None:
             cmds = sorted(cmds)
             for c in cmds:
                 display_cmd = c
@@ -955,7 +998,7 b' def help_('
                 )
         return rst
 
-    def helptopic(name, subtopic=None):
+    def helptopic(name: bytes, subtopic: Optional[bytes] = None) -> List[bytes]:
         # Look for sub-topic entry first.
         header, doc = None, None
         if subtopic and name in subtopics:
@@ -998,7 +1041,7 b' def help_('
             pass
         return rst
 
-    def helpext(name, subtopic=None):
+    def helpext(name: bytes, subtopic: Optional[bytes] = None) -> List[bytes]:
         try:
             mod = extensions.find(name)
             doc = gettext(pycompat.getdoc(mod)) or _(b'no help text available')
@@ -1040,7 +1083,9 b' def help_('
             )
         return rst
 
-    def helpextcmd(name, subtopic=None):
+    def helpextcmd(
+        name: bytes, subtopic: Optional[bytes] = None
+    ) -> List[bytes]:
         cmd, ext, doc = extensions.disabledcmd(
             ui, name, ui.configbool(b'ui', b'strict')
         )
@@ -1127,8 +1172,14 b' def help_('
 
 
 def formattedhelp(
-    ui, commands, fullname, keep=None, unknowncmd=False, full=True, **opts
-):
+    ui: uimod.ui,
+    commands,
+    fullname: Optional[bytes],
+    keep: Optional[Iterable[bytes]] = None,
+    unknowncmd: bool = False,
+    full: bool = True,
+    **opts
+) -> bytes:
     """get help for a given topic (as a dotted name) as rendered rst
 
     Either returns the rendered help text or raises an exception.

mercurial/helptext/config.txt

@@ -1922,6 +1922,42 b' The following sub-options can be defined'
   - ``ignore``: ignore bookmarks during exchange.
     (This currently only affect pulling)
 
+.. container:: verbose
+
+  ``delta-reuse-policy``
+  Control the policy regarding deltas sent by the remote during pulls.
+
+  This is an advanced option that non-admin users should not need to understand
+  or set. This option can be used to speed up pulls from trusted central
+  servers, or to fix-up deltas from older servers.
+
+  It supports the following values:
+
+  - ``default``: use the policy defined by
+    `storage.revlog.reuse-external-delta-parent`,
+
+  - ``no-reuse``: start a new optimal delta search for each new revision we add
+    to the repository. The deltas from the server will be reused when the base
+    it applies to is tested (this can be frequent if that base is the one and
+    unique parent of that revision). This can significantly slowdown pulls but
+    will result in an optimized storage space if the remote peer is sending poor
+    quality deltas.
+
+  - ``try-base``: try to reuse the deltas from the remote peer as long as they
+    create a valid delta-chain in the local repository. This speeds up the
+    unbundling process, but can result in sub-optimal storage space if the
+    remote peer is sending poor quality deltas.
+
+  - ``forced``: the deltas from the peer will be reused in all cases, even if
+    the resulting delta-chain is "invalid". This setting will ensure the bundle
+    is applied at minimal CPU cost, but it can result in longer delta chains
+    being created on the client, making revisions potentially slower to access
+    in the future. If you think you need this option, you should make sure you
+    are also talking to the Mercurial developer community to get confirmation.
+
+  See `hg help config.storage.revlog.reuse-external-delta-parent` for a similar
+  global option. That option defines the behavior of `default`.
+
 The following special named paths exist:
 
 ``default``
@@ -2281,6 +2317,21 b' category impact performance and reposito'
     To fix affected revisions that already exist within the repository, one can
     use :hg:`debug-repair-issue-6528`.
 
+.. container:: verbose
+
+    ``revlog.delta-parent-search.candidate-group-chunk-size``
+        Tune the number of delta bases the storage will consider in the
+        same "round" of search. In some very rare cases, using a smaller value
+        might result in faster processing at the possible expense of storage
+        space, while using larger values might result in slower processing at the
+        possible benefit of storage space. A value of "0" means no limitation.
+
+        default: no limitation
+
+        This is unlikely that you'll have to tune this configuration. If you think
+        you do, consider talking with the mercurial developer community about your
+        repositories.
+
 ``revlog.optimize-delta-parent-choice``
     When storing a merge revision, both parents will be equally considered as
     a possible delta base. This results in better delta selection and improved

mercurial/helptext/rust.txt

@@ -76,8 +76,8 b' instructions on how to install from sour'
 MSRV
 ====
 
-The minimum supported Rust version is currently 1.48.0. The project's policy is
-to follow the version from Debian stable, to make the distributions' job easier.
+The minimum supported Rust version is currently 1.61.0. The project's policy is
+to follow the version from Debian testing, to make the distributions' job easier.
 
 rhg
 ===

mercurial/hg.py

@@ -65,28 +65,12 b' release = lock.release'
 sharedbookmarks = b'bookmarks'
 
 
-def _local(path):
-    path = util.expandpath(urlutil.urllocalpath(path))
-
-    try:
-        # we use os.stat() directly here instead of os.path.isfile()
-        # because the latter started returning `False` on invalid path
-        # exceptions starting in 3.8 and we care about handling
-        # invalid paths specially here.
-        st = os.stat(path)
-        isfile = stat.S_ISREG(st.st_mode)
-    except ValueError as e:
-        raise error.Abort(
-            _(b'invalid path %s: %s') % (path, stringutil.forcebytestr(e))
-        )
-    except OSError:
-        isfile = False
-
-    return isfile and bundlerepo or localrepo
-
-
 def addbranchrevs(lrepo, other, branches, revs):
-    peer = other.peer()  # a courtesy to callers using a localrepo for other
+    if util.safehasattr(other, 'peer'):
+        # a courtesy to callers using a localrepo for other
+        peer = other.peer()
+    else:
+        peer = other
     hashbranch, branches = branches
     if not hashbranch and not branches:
         x = revs or None
@@ -129,10 +113,47 b' def addbranchrevs(lrepo, other, branches'
     return revs, revs[0]
 
 
-schemes = {
+def _isfile(path):
+    try:
+        # we use os.stat() directly here instead of os.path.isfile()
+        # because the latter started returning `False` on invalid path
+        # exceptions starting in 3.8 and we care about handling
+        # invalid paths specially here.
+        st = os.stat(path)
+    except ValueError as e:
+        msg = stringutil.forcebytestr(e)
+        raise error.Abort(_(b'invalid path %s: %s') % (path, msg))
+    except OSError:
+        return False
+    else:
+        return stat.S_ISREG(st.st_mode)
+
+
+class LocalFactory:
+    """thin wrapper to dispatch between localrepo and bundle repo"""
+
+    @staticmethod
+    def islocal(path: bytes) -> bool:
+        path = util.expandpath(urlutil.urllocalpath(path))
+        return not _isfile(path)
+
+    @staticmethod
+    def instance(ui, path, *args, **kwargs):
+        path = util.expandpath(urlutil.urllocalpath(path))
+        if _isfile(path):
+            cls = bundlerepo
+        else:
+            cls = localrepo
+        return cls.instance(ui, path, *args, **kwargs)
+
+
+repo_schemes = {
     b'bundle': bundlerepo,
     b'union': unionrepo,
-    b'file': _local,
+    b'file': LocalFactory,
+}
+
+peer_schemes = {
     b'http': httppeer,
     b'https': httppeer,
     b'ssh': sshpeer,
@@ -140,27 +161,23 b' schemes = {'
 }
 
 
-def _peerlookup(path):
-    u = urlutil.url(path)
-    scheme = u.scheme or b'file'
-    thing = schemes.get(scheme) or schemes[b'file']
-    try:
-        return thing(path)
-    except TypeError:
-        # we can't test callable(thing) because 'thing' can be an unloaded
-        # module that implements __call__
-        if not util.safehasattr(thing, b'instance'):
-            raise
-        return thing
-
-
 def islocal(repo):
     '''return true if repo (or path pointing to repo) is local'''
     if isinstance(repo, bytes):
-        try:
-            return _peerlookup(repo).islocal(repo)
-        except AttributeError:
-            return False
+        u = urlutil.url(repo)
+        scheme = u.scheme or b'file'
+        if scheme in peer_schemes:
+            cls = peer_schemes[scheme]
+            cls.make_peer  # make sure we load the module
+        elif scheme in repo_schemes:
+            cls = repo_schemes[scheme]
+            cls.instance  # make sure we load the module
+        else:
+            cls = LocalFactory
+        if util.safehasattr(cls, 'islocal'):
+            return cls.islocal(repo)  # pytype: disable=module-attr
+        return False
+    repo.ui.deprecwarn(b"use obj.local() instead of islocal(obj)", b"6.4")
     return repo.local()
 
 
@@ -177,13 +194,7 b' def openpath(ui, path, sendaccept=True):'
 wirepeersetupfuncs = []
 
 
-def _peerorrepo(
-    ui, path, create=False, presetupfuncs=None, intents=None, createopts=None
-):
-    """return a repository object for the specified path"""
-    obj = _peerlookup(path).instance(
-        ui, path, create, intents=intents, createopts=createopts
-    )
+def _setup_repo_or_peer(ui, obj, presetupfuncs=None):
     ui = getattr(obj, "ui", ui)
     for f in presetupfuncs or []:
         f(ui, obj)
@@ -195,14 +206,12 b' def _peerorrepo('
             if hook:
                 with util.timedcm('reposetup %r', name) as stats:
                     hook(ui, obj)
-                ui.log(
-                    b'extension', b'  > reposetup for %s took %s\n', name, stats
-                )
+                msg = b'  > reposetup for %s took %s\n'
+                ui.log(b'extension', msg, name, stats)
     ui.log(b'extension', b'> all reposetup took %s\n', allreposetupstats)
     if not obj.local():
         for f in wirepeersetupfuncs:
             f(ui, obj)
-    return obj
 
 
 def repository(
@@ -214,28 +223,59 b' def repository('
     createopts=None,
 ):
     """return a repository object for the specified path"""
-    peer = _peerorrepo(
+    scheme = urlutil.url(path).scheme
+    if scheme is None:
+        scheme = b'file'
+    cls = repo_schemes.get(scheme)
+    if cls is None:
+        if scheme in peer_schemes:
+            raise error.Abort(_(b"repository '%s' is not local") % path)
+        cls = LocalFactory
+    repo = cls.instance(
         ui,
         path,
         create,
-        presetupfuncs=presetupfuncs,
         intents=intents,
         createopts=createopts,
     )
-    repo = peer.local()
-    if not repo:
-        raise error.Abort(
-            _(b"repository '%s' is not local") % (path or peer.url())
-        )
+    _setup_repo_or_peer(ui, repo, presetupfuncs=presetupfuncs)
     return repo.filtered(b'visible')
 
 
 def peer(uiorrepo, opts, path, create=False, intents=None, createopts=None):
     '''return a repository peer for the specified path'''
+    ui = getattr(uiorrepo, 'ui', uiorrepo)
     rui = remoteui(uiorrepo, opts)
-    return _peerorrepo(
-        rui, path, create, intents=intents, createopts=createopts
-    ).peer()
+    if util.safehasattr(path, 'url'):
+        # this is already a urlutil.path object
+        peer_path = path
+    else:
+        peer_path = urlutil.path(ui, None, rawloc=path, validate_path=False)
+    scheme = peer_path.url.scheme  # pytype: disable=attribute-error
+    if scheme in peer_schemes:
+        cls = peer_schemes[scheme]
+        peer = cls.make_peer(
+            rui,
+            peer_path,
+            create,
+            intents=intents,
+            createopts=createopts,
+        )
+        _setup_repo_or_peer(rui, peer)
+    else:
+        # this is a repository
+        repo_path = peer_path.loc  # pytype: disable=attribute-error
+        if not repo_path:
+            repo_path = peer_path.rawloc  # pytype: disable=attribute-error
+        repo = repository(
+            rui,
+            repo_path,
+            create,
+            intents=intents,
+            createopts=createopts,
+        )
+        peer = repo.peer(path=peer_path)
+    return peer
 
 
 def defaultdest(source):
@@ -290,17 +330,23 b' def share('
 ):
     '''create a shared repository'''
 
-    if not islocal(source):
-        raise error.Abort(_(b'can only share local repositories'))
+    not_local_msg = _(b'can only share local repositories')
+    if util.safehasattr(source, 'local'):
+        if source.local() is None:
+            raise error.Abort(not_local_msg)
+    elif not islocal(source):
+        # XXX why are we getting bytes here ?
+        raise error.Abort(not_local_msg)
 
     if not dest:
         dest = defaultdest(source)
     else:
-        dest = urlutil.get_clone_path(ui, dest)[1]
+        dest = urlutil.get_clone_path_obj(ui, dest).loc
 
     if isinstance(source, bytes):
-        origsource, source, branches = urlutil.get_clone_path(ui, source)
-        srcrepo = repository(ui, source)
+        source_path = urlutil.get_clone_path_obj(ui, source)
+        srcrepo = repository(ui, source_path.loc)
+        branches = (source_path.branch, [])
         rev, checkout = addbranchrevs(srcrepo, srcrepo, branches, None)
     else:
         srcrepo = source.local()
@@ -661,12 +707,23 b' def clone('
     """
 
     if isinstance(source, bytes):
-        src = urlutil.get_clone_path(ui, source, branch)
-        origsource, source, branches = src
-        srcpeer = peer(ui, peeropts, source)
+        src_path = urlutil.get_clone_path_obj(ui, source)
+        if src_path is None:
+            srcpeer = peer(ui, peeropts, b'')
+            origsource = source = b''
+            branches = (None, branch or [])
+        else:
+            srcpeer = peer(ui, peeropts, src_path)
+            origsource = src_path.rawloc
+            branches = (src_path.branch, branch or [])
+            source = src_path.loc
     else:
-        srcpeer = source.peer()  # in case we were called with a localrepo
+        if util.safehasattr(source, 'peer'):
+            srcpeer = source.peer()  # in case we were called with a localrepo
+        else:
+            srcpeer = source
         branches = (None, branch or [])
+        # XXX path: simply use the peer `path` object when this become available
         origsource = source = srcpeer.url()
     srclock = destlock = destwlock = cleandir = None
     destpeer = None
@@ -678,7 +735,11 b' def clone('
             if dest:
                 ui.status(_(b"destination directory: %s\n") % dest)
         else:
-            dest = urlutil.get_clone_path(ui, dest)[0]
+            dest_path = urlutil.get_clone_path_obj(ui, dest)
+            if dest_path is not None:
+                dest = dest_path.rawloc
+            else:
+                dest = b''
 
         dest = urlutil.urllocalpath(dest)
         source = urlutil.urllocalpath(source)
@@ -1271,23 +1332,28 b' def _incoming('
         msg %= len(srcs)
         raise error.Abort(msg)
     path = srcs[0]
-    source, branches = urlutil.parseurl(path.rawloc, opts.get(b'branch'))
-    if subpath is not None:
+    if subpath is None:
+        peer_path = path
+        url = path.loc
+    else:
+        # XXX path: we are losing the `path` object here. Keeping it would be
+        # valuable. For example as a "variant" as we do for pushes.
         subpath = urlutil.url(subpath)
         if subpath.isabs():
-            source = bytes(subpath)
+            peer_path = url = bytes(subpath)
         else:
-            p = urlutil.url(source)
+            p = urlutil.url(path.loc)
             if p.islocal():
                 normpath = os.path.normpath
             else:
                 normpath = posixpath.normpath
             p.path = normpath(b'%s/%s' % (p.path, subpath))
-            source = bytes(p)
-    other = peer(repo, opts, source)
+            peer_path = url = bytes(p)
+    other = peer(repo, opts, peer_path)
     cleanupfn = other.close
     try:
-        ui.status(_(b'comparing with %s\n') % urlutil.hidepassword(source))
+        ui.status(_(b'comparing with %s\n') % urlutil.hidepassword(url))
+        branches = (path.branch, opts.get(b'branch', []))
         revs, checkout = addbranchrevs(repo, other, branches, opts.get(b'rev'))
 
         if revs:
@@ -1346,7 +1412,7 b' def _outgoing(ui, repo, dests, opts, sub'
     out = set()
     others = []
     for path in urlutil.get_push_paths(repo, ui, dests):
-        dest = path.pushloc or path.loc
+        dest = path.loc
         if subpath is not None:
             subpath = urlutil.url(subpath)
             if subpath.isabs():

mercurial/hgweb/hgweb_mod.py

@@ -230,8 +230,9 b' class requestcontext:'
 
     def sendtemplate(self, name, **kwargs):
         """Helper function to send a response generated from a template."""
-        kwargs = pycompat.byteskwargs(kwargs)
-        self.res.setbodygen(self.tmpl.generate(name, kwargs))
+        if self.req.method != b'HEAD':
+            kwargs = pycompat.byteskwargs(kwargs)
+            self.res.setbodygen(self.tmpl.generate(name, kwargs))
         return self.res.sendresponse()
 
 

mercurial/hgweb/request.py

@@ -485,6 +485,7 b' class wsgiresponse:'
             self._bodybytes is None
             and self._bodygen is None
             and not self._bodywillwrite
+            and self._req.method != b'HEAD'
         ):
             raise error.ProgrammingError(b'response body not defined')
 
@@ -594,6 +595,8 b' class wsgiresponse:'
                 yield chunk
         elif self._bodywillwrite:
             self._bodywritefn = write
+        elif self._req.method == b'HEAD':
+            pass
         else:
             error.ProgrammingError(b'do not know how to send body')
 

mercurial/hgweb/server.py

@@ -151,6 +151,9 b' class _httprequesthandler(httpservermod.'
     def do_GET(self):
         self.do_POST()
 
+    def do_HEAD(self):
+        self.do_POST()
+
     def do_hgweb(self):
         self.sent_headers = False
         path, query = _splitURI(self.path)
@@ -246,7 +249,11 b' class _httprequesthandler(httpservermod.'
             self.send_header(*h)
             if h[0].lower() == 'content-length':
                 self.length = int(h[1])
-        if self.length is None and saved_status[0] != common.HTTP_NOT_MODIFIED:
+        if (
+            self.length is None
+            and saved_status[0] != common.HTTP_NOT_MODIFIED
+            and self.command != 'HEAD'
+        ):
             self._chunked = (
                 not self.close_connection and self.request_version == 'HTTP/1.1'
             )

mercurial/hgweb/webcommands.py

@@ -1299,6 +1299,9 b' def archive(web):'
             b'sendresponse() should not emit data if writing later'
         )
 
+    if web.req.method == b'HEAD':
+        return []
+
     bodyfh = web.res.getbodyfile()
 
     archival.archive(

mercurial/httppeer.py

@@ -382,8 +382,7 b' def parsev1commandresponse(ui, baseurl, '
 
 class httppeer(wireprotov1peer.wirepeer):
     def __init__(self, ui, path, url, opener, requestbuilder, caps):
-        self.ui = ui
-        self._path = path
+        super().__init__(ui, path=path)
         self._url = url
         self._caps = caps
         self.limitedarguments = caps is not None and b'httppostargs' not in caps
@@ -398,14 +397,11 b' class httppeer(wireprotov1peer.wirepeer)'
     # Begin of ipeerconnection interface.
 
     def url(self):
-        return self._path
+        return self.path.loc
 
     def local(self):
         return None
 
-    def peer(self):
-        return self
-
     def canpush(self):
         return True
 
@@ -605,14 +601,13 b' def makepeer(ui, path, opener=None, requ'
     ``requestbuilder`` is the type used for constructing HTTP requests.
     It exists as an argument so extensions can override the default.
     """
-    u = urlutil.url(path)
-    if u.query or u.fragment:
-        raise error.Abort(
-            _(b'unsupported URL component: "%s"') % (u.query or u.fragment)
-        )
+    if path.url.query or path.url.fragment:
+        msg = _(b'unsupported URL component: "%s"')
+        msg %= path.url.query or path.url.fragment
+        raise error.Abort(msg)
 
     # urllib cannot handle URLs with embedded user or passwd.
-    url, authinfo = u.authinfo()
+    url, authinfo = path.url.authinfo()
     ui.debug(b'using %s\n' % url)
 
     opener = opener or urlmod.opener(ui, authinfo)
@@ -624,11 +619,11 b' def makepeer(ui, path, opener=None, requ'
     )
 
 
-def instance(ui, path, create, intents=None, createopts=None):
+def make_peer(ui, path, create, intents=None, createopts=None):
     if create:
         raise error.Abort(_(b'cannot create new http repository'))
     try:
-        if path.startswith(b'https:') and not urlmod.has_https:
+        if path.url.scheme == b'https' and not urlmod.has_https:
             raise error.Abort(
                 _(b'Python support for SSL and HTTPS is not installed')
             )
@@ -638,7 +633,7 b' def instance(ui, path, create, intents=N'
         return inst
     except error.RepoError as httpexception:
         try:
-            r = statichttprepo.instance(ui, b"static-" + path, create)
+            r = statichttprepo.make_peer(ui, b"static-" + path.loc, create)
             ui.note(_(b'(falling back to static-http)\n'))
             return r
         except error.RepoError:

mercurial/interfaces/dirstate.py

@@ -12,6 +12,7 b' class idirstate(interfaceutil.Interface)'
         sparsematchfn,
         nodeconstants,
         use_dirstate_v2,
+        use_tracked_hint=False,
     ):
         """Create a new dirstate object.
 
@@ -23,6 +24,15 b' class idirstate(interfaceutil.Interface)'
     # TODO: all these private methods and attributes should be made
     # public or removed from the interface.
     _ignore = interfaceutil.Attribute("""Matcher for ignored files.""")
+    is_changing_any = interfaceutil.Attribute(
+        """True if any changes in progress."""
+    )
+    is_changing_parents = interfaceutil.Attribute(
+        """True if parents changes in progress."""
+    )
+    is_changing_files = interfaceutil.Attribute(
+        """True if file tracking changes in progress."""
+    )
 
     def _ignorefiles():
         """Return a list of files containing patterns to ignore."""
@@ -34,7 +44,7 b' class idirstate(interfaceutil.Interface)'
     _checkexec = interfaceutil.Attribute("""Callable for checking exec bits.""")
 
     @contextlib.contextmanager
-    def parentchange():
+    def changing_parents(repo):
         """Context manager for handling dirstate parents.
 
         If an exception occurs in the scope of the context manager,
@@ -42,16 +52,26 b' class idirstate(interfaceutil.Interface)'
         released.
         """
 
-    def pendingparentchange():
-        """Returns true if the dirstate is in the middle of a set of changes
-        that modify the dirstate parent.
+    @contextlib.contextmanager
+    def changing_files(repo):
+        """Context manager for handling dirstate files.
+
+        If an exception occurs in the scope of the context manager,
+        the incoherent dirstate won't be written when wlock is
+        released.
         """
 
     def hasdir(d):
         pass
 
     def flagfunc(buildfallback):
-        pass
+        """build a callable that returns flags associated with a filename
+
+        The information is extracted from three possible layers:
+        1. the file system if it supports the information
+        2. the "fallback" information stored in the dirstate if any
+        3. a more expensive mechanism inferring the flags from the parents.
+        """
 
     def getcwd():
         """Return the path from which a canonical path is calculated.
@@ -61,12 +81,12 b' class idirstate(interfaceutil.Interface)'
         used to get real file paths. Use vfs functions instead.
         """
 
+    def pathto(f, cwd=None):
+        pass
+
     def get_entry(path):
         """return a DirstateItem for the associated path"""
 
-    def pathto(f, cwd=None):
-        pass
-
     def __contains__(key):
         """Check if bytestring `key` is known to the dirstate."""
 
@@ -96,7 +116,7 b' class idirstate(interfaceutil.Interface)'
     def setparents(p1, p2=None):
         """Set dirstate parents to p1 and p2.
 
-        When moving from two parents to one, 'm' merged entries a
+        When moving from two parents to one, "merged" entries a
         adjusted to normal and previous copy records discarded and
         returned by the call.
 
@@ -147,7 +167,7 b' class idirstate(interfaceutil.Interface)'
         pass
 
     def identity():
-        """Return identity of dirstate it to detect changing in storage
+        """Return identity of dirstate itself to detect changing in storage
 
         If identity of previous dirstate is equal to this, writing
         changes based on the former dirstate out can keep consistency.
@@ -200,11 +220,7 b' class idirstate(interfaceutil.Interface)'
         return files in the dirstate (in whatever state) filtered by match
         """
 
-    def savebackup(tr, backupname):
-        '''Save current dirstate into backup file'''
-
-    def restorebackup(tr, backupname):
-        '''Restore dirstate by backup file'''
-
-    def clearbackup(tr, backupname):
-        '''Clear backup file'''
+    def verify(m1, m2, p1, narrow_matcher=None):
+        """
+        check the dirstate contents against the parent manifest and yield errors
+        """

mercurial/interfaces/repository.py

@@ -103,6 +103,7 b' class ipeerconnection(interfaceutil.Inte'
     """
 
     ui = interfaceutil.Attribute("""ui.ui instance""")
+    path = interfaceutil.Attribute("""a urlutil.path instance or None""")
 
     def url():
         """Returns a URL string representing this peer.
@@ -123,12 +124,6 b' class ipeerconnection(interfaceutil.Inte'
         can be used to interface with it. Otherwise returns ``None``.
         """
 
-    def peer():
-        """Returns an object conforming to this interface.
-
-        Most implementations will ``return self``.
-        """
-
     def canpush():
         """Returns a boolean indicating if this peer can be pushed to."""
 
@@ -393,6 +388,10 b' class peer:'
 
     limitedarguments = False
 
+    def __init__(self, ui, path=None):
+        self.ui = ui
+        self.path = path
+
     def capable(self, name):
         caps = self.capabilities()
         if name in caps:
@@ -1613,7 +1612,7 b' class ilocalrepositorymain(interfaceutil'
     def close():
         """Close the handle on this repository."""
 
-    def peer():
+    def peer(path=None):
         """Obtain an object conforming to the ``peer`` interface."""
 
     def unfiltered():

mercurial/localrepo.py

@@ -10,11 +10,16 b''
 import functools
 import os
 import random
+import re
 import sys
 import time
 import weakref
 
 from concurrent import futures
+from typing import (
+    Optional,
+)
+
 from .i18n import _
 from .node import (
     bin,
@@ -37,7 +42,6 b' from . import ('
     commit,
     context,
     dirstate,
-    dirstateguard,
     discovery,
     encoding,
     error,
@@ -96,6 +100,8 b' release = lockmod.release'
 urlerr = util.urlerr
 urlreq = util.urlreq
 
+RE_SKIP_DIRSTATE_ROLLBACK = re.compile(b"^(dirstate|narrowspec.dirstate).*")
+
 # set of (path, vfs-location) tuples. vfs-location is:
 # - 'plain for vfs relative paths
 # - '' for svfs relative paths
@@ -299,13 +305,12 b' class localcommandexecutor:'
 class localpeer(repository.peer):
     '''peer for a local repo; reflects only the most recent API'''
 
-    def __init__(self, repo, caps=None):
-        super(localpeer, self).__init__()
+    def __init__(self, repo, caps=None, path=None):
+        super(localpeer, self).__init__(repo.ui, path=path)
 
         if caps is None:
             caps = moderncaps.copy()
         self._repo = repo.filtered(b'served')
-        self.ui = repo.ui
 
         if repo._wanted_sidedata:
             formatted = bundle2.format_remote_wanted_sidedata(repo)
@@ -321,9 +326,6 b' class localpeer(repository.peer):'
     def local(self):
         return self._repo
 
-    def peer(self):
-        return self
-
     def canpush(self):
         return True
 
@@ -451,8 +453,8 b' class locallegacypeer(localpeer):'
     """peer extension which implements legacy methods too; used for tests with
     restricted capabilities"""
 
-    def __init__(self, repo):
-        super(locallegacypeer, self).__init__(repo, caps=legacycaps)
+    def __init__(self, repo, path=None):
+        super(locallegacypeer, self).__init__(repo, caps=legacycaps, path=path)
 
     # Begin of baselegacywirecommands interface.
 
@@ -526,7 +528,7 b' def _readrequires(vfs, allowmissing):'
     return set(read(b'requires').splitlines())
 
 
-def makelocalrepository(baseui, path, intents=None):
+def makelocalrepository(baseui, path: bytes, intents=None):
     """Create a local repository object.
 
     Given arguments needed to construct a local repository, this function
@@ -612,7 +614,6 b' def makelocalrepository(baseui, path, in'
     # to be reshared
     hint = _(b"see `hg help config.format.use-share-safe` for more information")
     if requirementsmod.SHARESAFE_REQUIREMENT in requirements:
-
         if (
             shared
             and requirementsmod.SHARESAFE_REQUIREMENT
@@ -845,7 +846,13 b' def makelocalrepository(baseui, path, in'
     )
 
 
-def loadhgrc(ui, wdirvfs, hgvfs, requirements, sharedvfs=None):
+def loadhgrc(
+    ui,
+    wdirvfs: vfsmod.vfs,
+    hgvfs: vfsmod.vfs,
+    requirements,
+    sharedvfs: Optional[vfsmod.vfs] = None,
+):
     """Load hgrc files/content into a ui instance.
 
     This is called during repository opening to load any additional
@@ -1058,6 +1065,8 b' def resolverevlogstorevfsoptions(ui, req'
         options[b'revlogv2'] = True
     if requirementsmod.CHANGELOGV2_REQUIREMENT in requirements:
         options[b'changelogv2'] = True
+        cmp_rank = ui.configbool(b'experimental', b'changelog-v2.compute-rank')
+        options[b'changelogv2.compute-rank'] = cmp_rank
 
     if requirementsmod.GENERALDELTA_REQUIREMENT in requirements:
         options[b'generaldelta'] = True
@@ -1071,6 +1080,11 b' def resolverevlogstorevfsoptions(ui, req'
         b'storage', b'revlog.optimize-delta-parent-choice'
     )
     options[b'deltabothparents'] = deltabothparents
+    dps_cgds = ui.configint(
+        b'storage',
+        b'revlog.delta-parent-search.candidate-group-chunk-size',
+    )
+    options[b'delta-parent-search.candidate-group-chunk-size'] = dps_cgds
     options[b'debug-delta'] = ui.configbool(b'debug', b'revlog.debug-delta')
 
     issue6528 = ui.configbool(b'storage', b'revlog.issue6528.fix-incoming')
@@ -1311,8 +1325,6 b' class localrepository:'
         # XXX cache is a complicatged business someone
         # should investigate this in depth at some point
         b'cache/',
-        # XXX shouldn't be dirstate covered by the wlock?
-        b'dirstate',
         # XXX bisect was still a bit too messy at the time
         # this changeset was introduced. Someone should fix
         # the remainig bit and drop this line
@@ -1323,15 +1335,15 b' class localrepository:'
         self,
         baseui,
         ui,
-        origroot,
-        wdirvfs,
-        hgvfs,
+        origroot: bytes,
+        wdirvfs: vfsmod.vfs,
+        hgvfs: vfsmod.vfs,
         requirements,
         supportedrequirements,
-        sharedpath,
+        sharedpath: bytes,
         store,
-        cachevfs,
-        wcachevfs,
+        cachevfs: vfsmod.vfs,
+        wcachevfs: vfsmod.vfs,
         features,
         intents=None,
     ):
@@ -1453,6 +1465,7 b' class localrepository:'
         # - bookmark changes
         self.filteredrevcache = {}
 
+        self._dirstate = None
         # post-dirstate-status hooks
         self._postdsstatus = []
 
@@ -1620,8 +1633,8 b' class localrepository:'
                 parts.pop()
         return False
 
-    def peer(self):
-        return localpeer(self)  # not cached to avoid reference cycle
+    def peer(self, path=None):
+        return localpeer(self, path=path)  # not cached to avoid reference cycle
 
     def unfiltered(self):
         """Return unfiltered version of the repository
@@ -1738,9 +1751,13 b' class localrepository:'
     def manifestlog(self):
         return self.store.manifestlog(self, self._storenarrowmatch)
 
-    @repofilecache(b'dirstate')
+    @unfilteredpropertycache
     def dirstate(self):
-        return self._makedirstate()
+        if self._dirstate is None:
+            self._dirstate = self._makedirstate()
+        else:
+            self._dirstate.refresh()
+        return self._dirstate
 
     def _makedirstate(self):
         """Extension point for wrapping the dirstate per-repo."""
@@ -1977,7 +1994,7 b' class localrepository:'
     def __iter__(self):
         return iter(self.changelog)
 
-    def revs(self, expr, *args):
+    def revs(self, expr: bytes, *args):
         """Find revisions matching a revset.
 
         The revset is specified as a string ``expr`` that may contain
@@ -1993,7 +2010,7 b' class localrepository:'
         tree = revsetlang.spectree(expr, *args)
         return revset.makematcher(tree)(self)
 
-    def set(self, expr, *args):
+    def set(self, expr: bytes, *args):
         """Find revisions matching a revset and emit changectx instances.
 
         This is a convenience wrapper around ``revs()`` that iterates the
@@ -2005,7 +2022,7 b' class localrepository:'
         for r in self.revs(expr, *args):
             yield self[r]
 
-    def anyrevs(self, specs, user=False, localalias=None):
+    def anyrevs(self, specs: bytes, user=False, localalias=None):
         """Find revisions matching one of the given revsets.
 
         Revset aliases from the configuration are not expanded by default. To
@@ -2030,7 +2047,7 b' class localrepository:'
             m = revset.matchany(None, specs, localalias=localalias)
         return m(self)
 
-    def url(self):
+    def url(self) -> bytes:
         return b'file:' + self.root
 
     def hook(self, name, throw=False, **args):
@@ -2108,7 +2125,7 b' class localrepository:'
         # writing to the cache), but the rest of Mercurial wants them in
         # local encoding.
         tags = {}
-        for (name, (node, hist)) in alltags.items():
+        for name, (node, hist) in alltags.items():
             if node != self.nullid:
                 tags[encoding.tolocal(name)] = node
         tags[b'tip'] = self.changelog.tip()
@@ -2229,7 +2246,7 b' class localrepository:'
             return b'store'
         return None
 
-    def wjoin(self, f, *insidef):
+    def wjoin(self, f: bytes, *insidef: bytes) -> bytes:
         return self.vfs.reljoin(self.root, f, *insidef)
 
     def setparents(self, p1, p2=None):
@@ -2238,17 +2255,17 b' class localrepository:'
         self[None].setparents(p1, p2)
         self._quick_access_changeid_invalidate()
 
-    def filectx(self, path, changeid=None, fileid=None, changectx=None):
+    def filectx(self, path: bytes, changeid=None, fileid=None, changectx=None):
         """changeid must be a changeset revision, if specified.
         fileid can be a file revision or node."""
         return context.filectx(
             self, path, changeid, fileid, changectx=changectx
         )
 
-    def getcwd(self):
+    def getcwd(self) -> bytes:
         return self.dirstate.getcwd()
 
-    def pathto(self, f, cwd=None):
+    def pathto(self, f: bytes, cwd: Optional[bytes] = None) -> bytes:
         return self.dirstate.pathto(f, cwd)
 
     def _loadfilter(self, filter):
@@ -2300,14 +2317,21 b' class localrepository:'
     def adddatafilter(self, name, filter):
         self._datafilters[name] = filter
 
-    def wread(self, filename):
+    def wread(self, filename: bytes) -> bytes:
         if self.wvfs.islink(filename):
             data = self.wvfs.readlink(filename)
         else:
             data = self.wvfs.read(filename)
         return self._filter(self._encodefilterpats, filename, data)
 
-    def wwrite(self, filename, data, flags, backgroundclose=False, **kwargs):
+    def wwrite(
+        self,
+        filename: bytes,
+        data: bytes,
+        flags: bytes,
+        backgroundclose=False,
+        **kwargs
+    ) -> int:
         """write ``data`` into ``filename`` in the working directory
 
         This returns length of written (maybe decoded) data.
@@ -2325,7 +2349,7 b' class localrepository:'
                 self.wvfs.setflags(filename, False, False)
         return len(data)
 
-    def wwritedata(self, filename, data):
+    def wwritedata(self, filename: bytes, data: bytes) -> bytes:
         return self._filter(self._decodefilterpats, filename, data)
 
     def currenttransaction(self):
@@ -2356,6 +2380,21 b' class localrepository:'
                 hint=_(b"run 'hg recover' to clean up transaction"),
             )
 
+        # At that point your dirstate should be clean:
+        #
+        # - If you don't have the wlock, why would you still have a dirty
+        #   dirstate ?
+        #
+        # - If you hold the wlock, you should not be opening a transaction in
+        #   the middle of a `distate.changing_*` block. The transaction needs to
+        #   be open before that and wrap the change-context.
+        #
+        # - If you are not within a `dirstate.changing_*` context, why is our
+        #   dirstate dirty?
+        if self.dirstate._dirty:
+            m = "cannot open a transaction with a dirty dirstate"
+            raise error.ProgrammingError(m)
+
         idbase = b"%.40f#%f" % (random.random(), time.time())
         ha = hex(hashutil.sha1(idbase).digest())
         txnid = b'TXN:' + ha
@@ -2514,7 +2553,6 b' class localrepository:'
                 # out) in this transaction
                 narrowspec.restorebackup(self, b'journal.narrowspec')
                 narrowspec.restorewcbackup(self, b'journal.narrowspec.dirstate')
-                repo.dirstate.restorebackup(None, b'journal.dirstate')
 
                 repo.invalidate(clearfilecache=True)
 
@@ -2612,33 +2650,50 b' class localrepository:'
         tr.addpostclose(b'refresh-filecachestats', self._refreshfilecachestats)
         self._transref = weakref.ref(tr)
         scmutil.registersummarycallback(self, tr, desc)
+        # This only exist to deal with the need of rollback to have viable
+        # parents at the end of the operation. So backup viable parents at the
+        # time of this operation.
+        #
+        # We only do it when the `wlock` is taken, otherwise other might be
+        # altering the dirstate under us.
+        #
+        # This is really not a great way to do this (first, because we cannot
+        # always do it). There are more viable alternative that exists
+        #
+        # - backing only the working copy parent in a dedicated files and doing
+        #   a clean "keep-update" to them on `hg rollback`.
+        #
+        # - slightly changing the behavior an applying a logic similar to "hg
+        # strip" to pick a working copy destination on `hg rollback`
+        if self.currentwlock() is not None:
+            ds = self.dirstate
+
+            def backup_dirstate(tr):
+                for f in ds.all_file_names():
+                    # hardlink backup is okay because `dirstate` is always
+                    # atomically written and possible data file are append only
+                    # and resistant to trailing data.
+                    tr.addbackup(f, hardlink=True, location=b'plain')
+
+            tr.addvalidator(b'dirstate-backup', backup_dirstate)
         return tr
 
     def _journalfiles(self):
-        first = (
+        return (
             (self.svfs, b'journal'),
             (self.svfs, b'journal.narrowspec'),
             (self.vfs, b'journal.narrowspec.dirstate'),
-            (self.vfs, b'journal.dirstate'),
-        )
-        middle = []
-        dirstate_data = self.dirstate.data_backup_filename(b'journal.dirstate')
-        if dirstate_data is not None:
-            middle.append((self.vfs, dirstate_data))
-        end = (
             (self.vfs, b'journal.branch'),
             (self.vfs, b'journal.desc'),
             (bookmarks.bookmarksvfs(self), b'journal.bookmarks'),
             (self.svfs, b'journal.phaseroots'),
         )
-        return first + tuple(middle) + end
 
     def undofiles(self):
         return [(vfs, undoname(x)) for vfs, x in self._journalfiles()]
 
     @unfilteredmethod
     def _writejournal(self, desc):
-        self.dirstate.savebackup(None, b'journal.dirstate')
         narrowspec.savewcbackup(self, b'journal.narrowspec.dirstate')
         narrowspec.savebackup(self, b'journal.narrowspec')
         self.vfs.write(
@@ -2673,23 +2728,23 b' class localrepository:'
                 return False
 
     def rollback(self, dryrun=False, force=False):
-        wlock = lock = dsguard = None
+        wlock = lock = None
         try:
             wlock = self.wlock()
             lock = self.lock()
             if self.svfs.exists(b"undo"):
-                dsguard = dirstateguard.dirstateguard(self, b'rollback')
-
-                return self._rollback(dryrun, force, dsguard)
+                return self._rollback(dryrun, force)
             else:
                 self.ui.warn(_(b"no rollback information available\n"))
                 return 1
         finally:
-            release(dsguard, lock, wlock)
+            release(lock, wlock)
 
     @unfilteredmethod  # Until we get smarter cache management
-    def _rollback(self, dryrun, force, dsguard):
+    def _rollback(self, dryrun, force):
         ui = self.ui
+
+        parents = self.dirstate.parents()
         try:
             args = self.vfs.read(b'undo.desc').splitlines()
             (oldlen, desc, detail) = (int(args[0]), args[1], None)
@@ -2706,9 +2761,11 b' class localrepository:'
                 msg = _(
                     b'repository tip rolled back to revision %d (undo %s)\n'
                 ) % (oldtip, desc)
+            parentgone = any(self[p].rev() > oldtip for p in parents)
         except IOError:
             msg = _(b'rolling back unknown transaction\n')
             desc = None
+            parentgone = True
 
         if not force and self[b'.'] != self[b'tip'] and desc == b'commit':
             raise error.Abort(
@@ -2723,11 +2780,18 b' class localrepository:'
         if dryrun:
             return 0
 
-        parents = self.dirstate.parents()
         self.destroying()
         vfsmap = {b'plain': self.vfs, b'': self.svfs}
+        skip_journal_pattern = None
+        if not parentgone:
+            skip_journal_pattern = RE_SKIP_DIRSTATE_ROLLBACK
         transaction.rollback(
-            self.svfs, vfsmap, b'undo', ui.warn, checkambigfiles=_cachedfiles
+            self.svfs,
+            vfsmap,
+            b'undo',
+            ui.warn,
+            checkambigfiles=_cachedfiles,
+            skip_journal_pattern=skip_journal_pattern,
         )
         bookmarksvfs = bookmarks.bookmarksvfs(self)
         if bookmarksvfs.exists(b'undo.bookmarks'):
@@ -2737,16 +2801,20 b' class localrepository:'
         if self.svfs.exists(b'undo.phaseroots'):
             self.svfs.rename(b'undo.phaseroots', b'phaseroots', checkambig=True)
         self.invalidate()
-
-        has_node = self.changelog.index.has_node
-        parentgone = any(not has_node(p) for p in parents)
+        self.dirstate.invalidate()
+
         if parentgone:
-            # prevent dirstateguard from overwriting already restored one
-            dsguard.close()
+            # replace this with some explicit parent update in the future.
+            has_node = self.changelog.index.has_node
+            if not all(has_node(p) for p in self.dirstate._pl):
+                # There was no dirstate to backup initially, we need to drop
+                # the existing one.
+                with self.dirstate.changing_parents(self):
+                    self.dirstate.setparents(self.nullid)
+                    self.dirstate.clear()
 
             narrowspec.restorebackup(self, b'undo.narrowspec')
             narrowspec.restorewcbackup(self, b'undo.narrowspec.dirstate')
-            self.dirstate.restorebackup(None, b'undo.dirstate')
             try:
                 branch = self.vfs.read(b'undo.branch')
                 self.dirstate.setbranch(encoding.tolocal(branch))
@@ -2880,7 +2948,6 b' class localrepository:'
                 filtered.branchmap().write(filtered)
 
     def invalidatecaches(self):
-
         if '_tagscache' in vars(self):
             # can't use delattr on proxy
             del self.__dict__['_tagscache']
@@ -2903,13 +2970,9 b' class localrepository:'
         rereads the dirstate. Use dirstate.invalidate() if you want to
         explicitly read the dirstate again (i.e. restoring it to a previous
         known good state)."""
-        if hasunfilteredcache(self, 'dirstate'):
-            for k in self.dirstate._filecache:
-                try:
-                    delattr(self.dirstate, k)
-                except AttributeError:
-                    pass
-            delattr(self.unfiltered(), 'dirstate')
+        unfi = self.unfiltered()
+        if 'dirstate' in unfi.__dict__:
+            del unfi.__dict__['dirstate']
 
     def invalidate(self, clearfilecache=False):
         """Invalidates both store and non-store parts other than dirstate
@@ -2921,9 +2984,6 b' class localrepository:'
         """
         unfiltered = self.unfiltered()  # all file caches are stored unfiltered
         for k in list(self._filecache.keys()):
-            # dirstate is invalidated separately in invalidatedirstate()
-            if k == b'dirstate':
-                continue
             if (
                 k == b'changelog'
                 and self.currenttransaction()
@@ -3052,12 +3112,19 b' class localrepository:'
                 self.ui.develwarn(b'"wlock" acquired after "lock"')
 
         def unlock():
-            if self.dirstate.pendingparentchange():
+            if self.dirstate.is_changing_any:
+                msg = b"wlock release in the middle of a changing parents"
+                self.ui.develwarn(msg)
                 self.dirstate.invalidate()
             else:
+                if self.dirstate._dirty:
+                    msg = b"dirty dirstate on wlock release"
+                    self.ui.develwarn(msg)
                 self.dirstate.write(None)
 
-            self._filecache[b'dirstate'].refresh()
+            unfi = self.unfiltered()
+            if 'dirstate' in unfi.__dict__:
+                del unfi.__dict__['dirstate']
 
         l = self._lock(
             self.vfs,
@@ -3520,14 +3587,13 b' def aftertrans(files):'
     return a
 
 
-def undoname(fn):
+def undoname(fn: bytes) -> bytes:
     base, name = os.path.split(fn)
     assert name.startswith(b'journal')
     return os.path.join(base, name.replace(b'journal', b'undo', 1))
 
 
-def instance(ui, path, create, intents=None, createopts=None):
-
+def instance(ui, path: bytes, create, intents=None, createopts=None):
     # prevent cyclic import localrepo -> upgrade -> localrepo
     from . import upgrade
 
@@ -3543,7 +3609,7 b' def instance(ui, path, create, intents=N'
     return repo
 
 
-def islocal(path):
+def islocal(path: bytes) -> bool:
     return True
 
 
@@ -3803,7 +3869,7 b' def filterknowncreateopts(ui, createopts'
     return {k: v for k, v in createopts.items() if k not in known}
 
 
-def createrepository(ui, path, createopts=None, requirements=None):
+def createrepository(ui, path: bytes, createopts=None, requirements=None):
     """Create a new repository in a vfs.
 
     ``path`` path to the new repo's working directory.

mercurial/logexchange.py

@@ -113,7 +113,7 b' def activepath(repo, remote):'
     if local:
         rpath = util.pconvert(remote._repo.root)
     elif not isinstance(remote, bytes):
-        rpath = remote._url
+        rpath = remote.url()
 
     # represent the remotepath with user defined path name if exists
     for path, url in repo.ui.configitems(b'paths'):

mercurial/manifest.py

@@ -1836,6 +1836,7 b' class manifestrevlog:'
         assumehaveparentrevisions=False,
         deltamode=repository.CG_DELTAMODE_STD,
         sidedata_helpers=None,
+        debug_info=None,
     ):
         return self._revlog.emitrevisions(
             nodes,
@@ -1844,6 +1845,7 b' class manifestrevlog:'
             assumehaveparentrevisions=assumehaveparentrevisions,
             deltamode=deltamode,
             sidedata_helpers=sidedata_helpers,
+            debug_info=debug_info,
         )
 
     def addgroup(
@@ -1854,6 +1856,8 b' class manifestrevlog:'
         alwayscache=False,
         addrevisioncb=None,
         duplicaterevisioncb=None,
+        debug_info=None,
+        delta_base_reuse_policy=None,
     ):
         return self._revlog.addgroup(
             deltas,
@@ -1862,6 +1866,8 b' class manifestrevlog:'
             alwayscache=alwayscache,
             addrevisioncb=addrevisioncb,
             duplicaterevisioncb=duplicaterevisioncb,
+            debug_info=debug_info,
+            delta_base_reuse_policy=delta_base_reuse_policy,
         )
 
     def rawsize(self, rev):

mercurial/match.py

@@ -368,7 +368,7 b' def _donormalize(patterns, default, root'
                     % (
                         pat,
                         inst.message,
-                    )  # pytype: disable=unsupported-operands
+                    )
                 )
             except IOError as inst:
                 if warn:

mercurial/mdiff.py

@@ -94,6 +94,13 b' class diffopts:'
         opts.update(kwargs)
         return diffopts(**opts)
 
+    def __bytes__(self):
+        return b", ".join(
+            b"%s: %r" % (k, getattr(self, k)) for k in self.defaults
+        )
+
+    __str__ = encoding.strmethod(__bytes__)
+
 
 defaultopts = diffopts()
 

mercurial/merge.py

@@ -46,7 +46,7 b' def _getcheckunknownconfig(repo, section'
     return config
 
 
-def _checkunknownfile(repo, wctx, mctx, f, f2=None):
+def _checkunknownfile(dirstate, wvfs, dircache, wctx, mctx, f, f2=None):
     if wctx.isinmemory():
         # Nothing to do in IMM because nothing in the "working copy" can be an
         # unknown file.
@@ -58,9 +58,8 b' def _checkunknownfile(repo, wctx, mctx, '
     if f2 is None:
         f2 = f
     return (
-        repo.wvfs.audit.check(f)
-        and repo.wvfs.isfileorlink(f)
-        and repo.dirstate.normalize(f) not in repo.dirstate
+        wvfs.isfileorlink_checkdir(dircache, f)
+        and dirstate.normalize(f) not in dirstate
         and mctx[f2].cmp(wctx[f])
     )
 
@@ -136,6 +135,9 b' def _checkunknownfiles(repo, wctx, mctx,'
     pathconfig = repo.ui.configbool(
         b'experimental', b'merge.checkpathconflicts'
     )
+    dircache = dict()
+    dirstate = repo.dirstate
+    wvfs = repo.wvfs
     if not force:
 
         def collectconflicts(conflicts, config):
@@ -151,7 +153,7 b' def _checkunknownfiles(repo, wctx, mctx,'
                 mergestatemod.ACTION_DELETED_CHANGED,
             )
         ):
-            if _checkunknownfile(repo, wctx, mctx, f):
+            if _checkunknownfile(dirstate, wvfs, dircache, wctx, mctx, f):
                 fileconflicts.add(f)
             elif pathconfig and f not in wctx:
                 path = checkunknowndirs(repo, wctx, f)
@@ -160,7 +162,9 b' def _checkunknownfiles(repo, wctx, mctx,'
         for f, args, msg in mresult.getactions(
             [mergestatemod.ACTION_LOCAL_DIR_RENAME_GET]
         ):
-            if _checkunknownfile(repo, wctx, mctx, f, args[0]):
+            if _checkunknownfile(
+                dirstate, wvfs, dircache, wctx, mctx, f, args[0]
+            ):
                 fileconflicts.add(f)
 
         allconflicts = fileconflicts | pathconflicts
@@ -173,7 +177,9 b' def _checkunknownfiles(repo, wctx, mctx,'
             mresult.getactions([mergestatemod.ACTION_CREATED_MERGE])
         ):
             fl2, anc = args
-            different = _checkunknownfile(repo, wctx, mctx, f)
+            different = _checkunknownfile(
+                dirstate, wvfs, dircache, wctx, mctx, f
+            )
             if repo.dirstate._ignore(f):
                 config = ignoredconfig
             else:
@@ -240,16 +246,21 b' def _checkunknownfiles(repo, wctx, mctx,'
         else:
             repo.ui.warn(_(b"%s: replacing untracked files in directory\n") % f)
 
-    for f, args, msg in list(
-        mresult.getactions([mergestatemod.ACTION_CREATED])
-    ):
+    def transformargs(f, args):
         backup = (
             f in fileconflicts
-            or f in pathconflicts
-            or any(p in pathconflicts for p in pathutil.finddirs(f))
+            or pathconflicts
+            and (
+                f in pathconflicts
+                or any(p in pathconflicts for p in pathutil.finddirs(f))
+            )
         )
         (flags,) = args
-        mresult.addfile(f, mergestatemod.ACTION_GET, (flags, backup), msg)
+        return (flags, backup)
+
+    mresult.mapaction(
+        mergestatemod.ACTION_CREATED, mergestatemod.ACTION_GET, transformargs
+    )
 
 
 def _forgetremoved(wctx, mctx, branchmerge, mresult):
@@ -581,6 +592,18 b' class mergeresult:'
         self._filemapping[filename] = (action, data, message)
         self._actionmapping[action][filename] = (data, message)
 
+    def mapaction(self, actionfrom, actionto, transform):
+        """changes all occurrences of action `actionfrom` into `actionto`,
+        transforming its args with the function `transform`.
+        """
+        orig = self._actionmapping[actionfrom]
+        del self._actionmapping[actionfrom]
+        dest = self._actionmapping[actionto]
+        for f, (data, msg) in orig.items():
+            data = transform(f, data)
+            self._filemapping[f] = (actionto, data, msg)
+            dest[f] = (data, msg)
+
     def getfile(self, filename, default_return=None):
         """returns (action, args, msg) about this file
 
@@ -1142,6 +1165,8 b' def calculateupdates('
             followcopies,
         )
         _checkunknownfiles(repo, wctx, mctx, force, mresult, mergeforce)
+        if repo.ui.configbool(b'devel', b'debug.abort-update'):
+            exit(1)
 
     else:  # only when merge.preferancestor=* - the default
         repo.ui.note(
@@ -2130,7 +2155,7 b' def _update('
             assert len(getfiledata) == (
                 mresult.len((mergestatemod.ACTION_GET,)) if wantfiledata else 0
             )
-            with repo.dirstate.parentchange():
+            with repo.dirstate.changing_parents(repo):
                 ### Filter Filedata
                 #
                 # We gathered "cache" information for the clean file while
@@ -2352,7 +2377,7 b' def graft('
         # fix up dirstate for copies and renames
         copies.graftcopies(wctx, ctx, base)
     else:
-        with repo.dirstate.parentchange():
+        with repo.dirstate.changing_parents(repo):
             repo.setparents(pctx.node(), pother)
             repo.dirstate.write(repo.currenttransaction())
             # fix up dirstate for copies and renames

mercurial/narrowspec.py

@@ -322,10 +322,16 b' def updateworkingcopy(repo, assumeclean='
     addedmatch = matchmod.differencematcher(newmatch, oldmatch)
     removedmatch = matchmod.differencematcher(oldmatch, newmatch)
 
+    assert repo.currentwlock() is not None
     ds = repo.dirstate
-    lookup, status, _mtime_boundary = ds.status(
-        removedmatch, subrepos=[], ignored=True, clean=True, unknown=True
-    )
+    with ds.running_status(repo):
+        lookup, status, _mtime_boundary = ds.status(
+            removedmatch,
+            subrepos=[],
+            ignored=True,
+            clean=True,
+            unknown=True,
+        )
     trackeddirty = status.modified + status.added
     clean = status.clean
     if assumeclean:

mercurial/patch.py

@@ -570,22 +570,23 b' class workingbackend(fsbackend):'
         self.changed.add(fname)
 
     def close(self):
-        wctx = self.repo[None]
-        changed = set(self.changed)
-        for src, dst in self.copied:
-            scmutil.dirstatecopy(self.ui, self.repo, wctx, src, dst)
-        if self.removed:
-            wctx.forget(sorted(self.removed))
-            for f in self.removed:
-                if f not in self.repo.dirstate:
-                    # File was deleted and no longer belongs to the
-                    # dirstate, it was probably marked added then
-                    # deleted, and should not be considered by
-                    # marktouched().
-                    changed.discard(f)
-        if changed:
-            scmutil.marktouched(self.repo, changed, self.similarity)
-        return sorted(self.changed)
+        with self.repo.dirstate.changing_files(self.repo):
+            wctx = self.repo[None]
+            changed = set(self.changed)
+            for src, dst in self.copied:
+                scmutil.dirstatecopy(self.ui, self.repo, wctx, src, dst)
+            if self.removed:
+                wctx.forget(sorted(self.removed))
+                for f in self.removed:
+                    if f not in self.repo.dirstate:
+                        # File was deleted and no longer belongs to the
+                        # dirstate, it was probably marked added then
+                        # deleted, and should not be considered by
+                        # marktouched().
+                        changed.discard(f)
+            if changed:
+                scmutil.marktouched(self.repo, changed, self.similarity)
+            return sorted(self.changed)
 
 
 class filestore:

mercurial/pathutil.py

@@ -4,6 +4,13 b' import os'
 import posixpath
 import stat
 
+from typing import (
+    Any,
+    Callable,
+    Iterator,
+    Optional,
+)
+
 from .i18n import _
 from . import (
     encoding,
@@ -13,15 +20,6 b' from . import ('
     util,
 )
 
-if pycompat.TYPE_CHECKING:
-    from typing import (
-        Any,
-        Callable,
-        Iterator,
-        Optional,
-    )
-
-
 rustdirs = policy.importrust('dirstate', 'Dirs')
 parsers = policy.importmod('parsers')
 
@@ -56,7 +54,7 b' class pathauditor:'
 
     def __init__(self, root, callback=None, realfs=True, cached=False):
         self.audited = set()
-        self.auditeddir = set()
+        self.auditeddir = dict()
         self.root = root
         self._realfs = realfs
         self._cached = cached
@@ -72,8 +70,7 b' class pathauditor:'
         path may contain a pattern (e.g. foodir/**.txt)"""
 
         path = util.localpath(path)
-        normpath = self.normcase(path)
-        if normpath in self.audited:
+        if path in self.audited:
             return
         # AIX ignores "/" at end of path, others raise EISDIR.
         if util.endswithsep(path):
@@ -90,13 +87,14 b' class pathauditor:'
                 _(b"path contains illegal component: %s") % path
             )
         # Windows shortname aliases
-        for p in parts:
-            if b"~" in p:
-                first, last = p.split(b"~", 1)
-                if last.isdigit() and first.upper() in [b"HG", b"HG8B6C"]:
-                    raise error.InputError(
-                        _(b"path contains illegal component: %s") % path
-                    )
+        if b"~" in path:
+            for p in parts:
+                if b"~" in p:
+                    first, last = p.split(b"~", 1)
+                    if last.isdigit() and first.upper() in [b"HG", b"HG8B6C"]:
+                        raise error.InputError(
+                            _(b"path contains illegal component: %s") % path
+                        )
         if b'.hg' in _lowerclean(path):
             lparts = [_lowerclean(p) for p in parts]
             for p in b'.hg', b'.hg.':
@@ -108,36 +106,43 b' class pathauditor:'
                         % (path, pycompat.bytestr(base))
                     )
 
-        normparts = util.splitpath(normpath)
-        assert len(parts) == len(normparts)
-
-        parts.pop()
-        normparts.pop()
-        # It's important that we check the path parts starting from the root.
-        # We don't want to add "foo/bar/baz" to auditeddir before checking if
-        # there's a "foo/.hg" directory. This also means we won't accidentally
-        # traverse a symlink into some other filesystem (which is potentially
-        # expensive to access).
-        for i in range(len(parts)):
-            prefix = pycompat.ossep.join(parts[: i + 1])
-            normprefix = pycompat.ossep.join(normparts[: i + 1])
-            if normprefix in self.auditeddir:
-                continue
-            if self._realfs:
-                self._checkfs(prefix, path)
-            if self._cached:
-                self.auditeddir.add(normprefix)
+        if self._realfs:
+            # It's important that we check the path parts starting from the root.
+            # We don't want to add "foo/bar/baz" to auditeddir before checking if
+            # there's a "foo/.hg" directory. This also means we won't accidentally
+            # traverse a symlink into some other filesystem (which is potentially
+            # expensive to access).
+            for prefix in finddirs_rev_noroot(path):
+                if prefix in self.auditeddir:
+                    res = self.auditeddir[prefix]
+                else:
+                    res = pathauditor._checkfs_exists(
+                        self.root, prefix, path, self.callback
+                    )
+                    if self._cached:
+                        self.auditeddir[prefix] = res
+                if not res:
+                    break
 
         if self._cached:
-            self.audited.add(normpath)
+            self.audited.add(path)
 
-    def _checkfs(self, prefix, path):
-        # type: (bytes, bytes) -> None
-        """raise exception if a file system backed check fails"""
-        curpath = os.path.join(self.root, prefix)
+    @staticmethod
+    def _checkfs_exists(
+        root,
+        prefix: bytes,
+        path: bytes,
+        callback: Optional[Callable[[bytes], bool]] = None,
+    ):
+        """raise exception if a file system backed check fails.
+
+        Return a bool that indicates that the directory (or file) exists."""
+        curpath = os.path.join(root, prefix)
         try:
             st = os.lstat(curpath)
         except OSError as err:
+            if err.errno == errno.ENOENT:
+                return False
             # EINVAL can be raised as invalid path syntax under win32.
             # They must be ignored for patterns can be checked too.
             if err.errno not in (errno.ENOENT, errno.ENOTDIR, errno.EINVAL):
@@ -152,9 +157,10 b' class pathauditor:'
             elif stat.S_ISDIR(st.st_mode) and os.path.isdir(
                 os.path.join(curpath, b'.hg')
             ):
-                if not self.callback or not self.callback(curpath):
+                if not callback or not callback(curpath):
                     msg = _(b"path '%s' is inside nested repo %r")
                     raise error.Abort(msg % (path, pycompat.bytestr(prefix)))
+        return True
 
     def check(self, path):
         # type: (bytes) -> bool
@@ -314,6 +320,13 b' def finddirs(path):'
     yield b''
 
 
+def finddirs_rev_noroot(path: bytes) -> Iterator[bytes]:
+    pos = path.find(pycompat.ossep)
+    while pos != -1:
+        yield path[:pos]
+        pos = path.find(pycompat.ossep, pos + 1)
+
+
 class dirs:
     '''a multiset of directory names from a set of file paths'''
 

mercurial/policy.py

@@ -76,7 +76,7 b' def _importfrom(pkgname, modname):'
     ('cext', 'bdiff'): 3,
     ('cext', 'mpatch'): 1,
     ('cext', 'osutil'): 4,
-    ('cext', 'parsers'): 20,
+    ('cext', 'parsers'): 21,
 }
 
 # map import request to other package or module

mercurial/posix.py

@@ -17,8 +17,23 b' import select'
 import stat
 import sys
 import tempfile
+import typing
 import unicodedata
 
+from typing import (
+    Any,
+    AnyStr,
+    Iterable,
+    Iterator,
+    List,
+    Match,
+    NoReturn,
+    Optional,
+    Sequence,
+    Tuple,
+    Union,
+)
+
 from .i18n import _
 from .pycompat import (
     getattr,
@@ -44,7 +59,7 b' except AttributeError:'
     # vaguely unix-like but don't have hardlink support. For those
     # poor souls, just say we tried and that it failed so we fall back
     # to copies.
-    def oslink(src, dst):
+    def oslink(src: bytes, dst: bytes) -> NoReturn:
         raise OSError(
             errno.EINVAL, b'hardlinks not supported: %s to %s' % (src, dst)
         )
@@ -54,15 +69,47 b' readlink = os.readlink'
 unlink = os.unlink
 rename = os.rename
 removedirs = os.removedirs
-expandglobs = False
+
+if typing.TYPE_CHECKING:
+    # Replace the various overloads that come along with aliasing stdlib methods
+    # with the narrow definition that we care about in the type checking phase
+    # only.  This ensures that both Windows and POSIX see only the definition
+    # that is actually available.
+    #
+    # Note that if we check pycompat.TYPE_CHECKING here, it is always False, and
+    # the methods aren't replaced.
+
+    def normpath(path: bytes) -> bytes:
+        raise NotImplementedError
+
+    def abspath(path: AnyStr) -> AnyStr:
+        raise NotImplementedError
 
-umask = os.umask(0)
+    def oslink(src: bytes, dst: bytes) -> None:
+        raise NotImplementedError
+
+    def readlink(path: bytes) -> bytes:
+        raise NotImplementedError
+
+    def unlink(path: bytes) -> None:
+        raise NotImplementedError
+
+    def rename(src: bytes, dst: bytes) -> None:
+        raise NotImplementedError
+
+    def removedirs(name: bytes) -> None:
+        raise NotImplementedError
+
+
+expandglobs: bool = False
+
+umask: int = os.umask(0)
 os.umask(umask)
 
 posixfile = open
 
 
-def split(p):
+def split(p: bytes) -> Tuple[bytes, bytes]:
     """Same as posixpath.split, but faster
 
     >>> import posixpath
@@ -85,17 +132,17 b' def split(p):'
     return ht[0] + b'/', ht[1]
 
 
-def openhardlinks():
+def openhardlinks() -> bool:
     '''return true if it is safe to hold open file handles to hardlinks'''
     return True
 
 
-def nlinks(name):
+def nlinks(name: bytes) -> int:
     '''return number of hardlinks for the given file'''
     return os.lstat(name).st_nlink
 
 
-def parsepatchoutput(output_line):
+def parsepatchoutput(output_line: bytes) -> bytes:
     """parses the output produced by patch and returns the filename"""
     pf = output_line[14:]
     if pycompat.sysplatform == b'OpenVMS':
@@ -107,7 +154,9 b' def parsepatchoutput(output_line):'
     return pf
 
 
-def sshargs(sshcmd, host, user, port):
+def sshargs(
+    sshcmd: bytes, host: bytes, user: Optional[bytes], port: Optional[bytes]
+) -> bytes:
     '''Build argument list for ssh'''
     args = user and (b"%s@%s" % (user, host)) or host
     if b'-' in args[:1]:
@@ -120,12 +169,12 b' def sshargs(sshcmd, host, user, port):'
     return args
 
 
-def isexec(f):
+def isexec(f: bytes) -> bool:
     """check whether a file is executable"""
     return os.lstat(f).st_mode & 0o100 != 0
 
 
-def setflags(f, l, x):
+def setflags(f: bytes, l: bool, x: bool) -> None:
     st = os.lstat(f)
     s = st.st_mode
     if l:
@@ -169,7 +218,12 b' def setflags(f, l, x):'
         os.chmod(f, s & 0o666)
 
 
-def copymode(src, dst, mode=None, enforcewritable=False):
+def copymode(
+    src: bytes,
+    dst: bytes,
+    mode: Optional[bytes] = None,
+    enforcewritable: bool = False,
+) -> None:
     """Copy the file mode from the file at path src to dst.
     If src doesn't exist, we're using mode instead. If mode is None, we're
     using umask."""
@@ -189,7 +243,7 b' def copymode(src, dst, mode=None, enforc'
     os.chmod(dst, new_mode)
 
 
-def checkexec(path):
+def checkexec(path: bytes) -> bool:
     """
     Check whether the given path is on a filesystem with UNIX-like exec flags
 
@@ -230,7 +284,7 b' def checkexec(path):'
             else:
                 # checkisexec exists, check if it actually is exec
                 if m & EXECFLAGS != 0:
-                    # ensure checkisexec exists, check it isn't exec
+                    # ensure checknoexec exists, check it isn't exec
                     try:
                         m = os.stat(checknoexec).st_mode
                     except FileNotFoundError:
@@ -269,7 +323,7 b' def checkexec(path):'
         return False
 
 
-def checklink(path):
+def checklink(path: bytes) -> bool:
     """check whether the given path is on a symlink-capable filesystem"""
     # mktemp is not racy because symlink creation will fail if the
     # file already exists
@@ -334,13 +388,13 b' def checklink(path):'
             return False
 
 
-def checkosfilename(path):
+def checkosfilename(path: bytes) -> Optional[bytes]:
     """Check that the base-relative path is a valid filename on this platform.
     Returns None if the path is ok, or a UI string describing the problem."""
     return None  # on posix platforms, every path is ok
 
 
-def getfsmountpoint(dirpath):
+def getfsmountpoint(dirpath: bytes) -> Optional[bytes]:
     """Get the filesystem mount point from a directory (best-effort)
 
     Returns None if we are unsure. Raises OSError on ENOENT, EPERM, etc.
@@ -348,7 +402,7 b' def getfsmountpoint(dirpath):'
     return getattr(osutil, 'getfsmountpoint', lambda x: None)(dirpath)
 
 
-def getfstype(dirpath):
+def getfstype(dirpath: bytes) -> Optional[bytes]:
     """Get the filesystem type name from a directory (best-effort)
 
     Returns None if we are unsure. Raises OSError on ENOENT, EPERM, etc.
@@ -356,29 +410,29 b' def getfstype(dirpath):'
     return getattr(osutil, 'getfstype', lambda x: None)(dirpath)
 
 
-def get_password():
+def get_password() -> bytes:
     return encoding.strtolocal(getpass.getpass(''))
 
 
-def setbinary(fd):
+def setbinary(fd) -> None:
     pass
 
 
-def pconvert(path):
+def pconvert(path: bytes) -> bytes:
     return path
 
 
-def localpath(path):
+def localpath(path: bytes) -> bytes:
     return path
 
 
-def samefile(fpath1, fpath2):
+def samefile(fpath1: bytes, fpath2: bytes) -> bool:
     """Returns whether path1 and path2 refer to the same file. This is only
     guaranteed to work for files, not directories."""
     return os.path.samefile(fpath1, fpath2)
 
 
-def samedevice(fpath1, fpath2):
+def samedevice(fpath1: bytes, fpath2: bytes) -> bool:
     """Returns whether fpath1 and fpath2 are on the same device. This is only
     guaranteed to work for files, not directories."""
     st1 = os.lstat(fpath1)
@@ -387,18 +441,18 b' def samedevice(fpath1, fpath2):'
 
 
 # os.path.normcase is a no-op, which doesn't help us on non-native filesystems
-def normcase(path):
+def normcase(path: bytes) -> bytes:
     return path.lower()
 
 
 # what normcase does to ASCII strings
-normcasespec = encoding.normcasespecs.lower
+normcasespec: int = encoding.normcasespecs.lower
 # fallback normcase function for non-ASCII strings
 normcasefallback = normcase
 
 if pycompat.isdarwin:
 
-    def normcase(path):
+    def normcase(path: bytes) -> bytes:
         """
         Normalize a filename for OS X-compatible comparison:
         - escape-encode invalid characters
@@ -423,7 +477,7 b' if pycompat.isdarwin:'
 
     normcasespec = encoding.normcasespecs.lower
 
-    def normcasefallback(path):
+    def normcasefallback(path: bytes) -> bytes:
         try:
             u = path.decode('utf-8')
         except UnicodeDecodeError:
@@ -464,7 +518,7 b" if pycompat.sysplatform == b'cygwin':"
     )
 
     # use upper-ing as normcase as same as NTFS workaround
-    def normcase(path):
+    def normcase(path: bytes) -> bytes:
         pathlen = len(path)
         if (pathlen == 0) or (path[0] != pycompat.ossep):
             # treat as relative
@@ -490,20 +544,20 b" if pycompat.sysplatform == b'cygwin':"
     # but these translations are not supported by native
     # tools, so the exec bit tends to be set erroneously.
     # Therefore, disable executable bit access on Cygwin.
-    def checkexec(path):
+    def checkexec(path: bytes) -> bool:
         return False
 
     # Similarly, Cygwin's symlink emulation is likely to create
     # problems when Mercurial is used from both Cygwin and native
     # Windows, with other native tools, or on shared volumes
-    def checklink(path):
+    def checklink(path: bytes) -> bool:
         return False
 
 
-_needsshellquote = None
+_needsshellquote: Optional[Match[bytes]] = None
 
 
-def shellquote(s):
+def shellquote(s: bytes) -> bytes:
     if pycompat.sysplatform == b'OpenVMS':
         return b'"%s"' % s
     global _needsshellquote
@@ -516,12 +570,12 b' def shellquote(s):'
         return b"'%s'" % s.replace(b"'", b"'\\''")
 
 
-def shellsplit(s):
+def shellsplit(s: bytes) -> List[bytes]:
     """Parse a command string in POSIX shell way (best-effort)"""
     return pycompat.shlexsplit(s, posix=True)
 
 
-def testpid(pid):
+def testpid(pid: int) -> bool:
     '''return False if pid dead, True if running or not sure'''
     if pycompat.sysplatform == b'OpenVMS':
         return True
@@ -532,12 +586,12 b' def testpid(pid):'
         return inst.errno != errno.ESRCH
 
 
-def isowner(st):
+def isowner(st: os.stat_result) -> bool:
     """Return True if the stat object st is from the current user."""
     return st.st_uid == os.getuid()
 
 
-def findexe(command):
+def findexe(command: bytes) -> Optional[bytes]:
     """Find executable for command searching like which does.
     If command is a basename then PATH is searched for command.
     PATH isn't searched if command is an absolute or relative path.
@@ -545,7 +599,7 b' def findexe(command):'
     if pycompat.sysplatform == b'OpenVMS':
         return command
 
-    def findexisting(executable):
+    def findexisting(executable: bytes) -> Optional[bytes]:
         b'Will return executable if existing file'
         if os.path.isfile(executable) and os.access(executable, os.X_OK):
             return executable
@@ -564,14 +618,14 b' def findexe(command):'
     return None
 
 
-def setsignalhandler():
+def setsignalhandler() -> None:
     pass
 
 
 _wantedkinds = {stat.S_IFREG, stat.S_IFLNK}
 
 
-def statfiles(files):
+def statfiles(files: Sequence[bytes]) -> Iterator[Optional[os.stat_result]]:
     """Stat each file in files. Yield each stat, or None if a file does not
     exist or has a type we don't care about."""
     lstat = os.lstat
@@ -586,12 +640,12 b' def statfiles(files):'
         yield st
 
 
-def getuser():
+def getuser() -> bytes:
     '''return name of current user'''
     return pycompat.fsencode(getpass.getuser())
 
 
-def username(uid=None):
+def username(uid: Optional[int] = None) -> Optional[bytes]:
     """Return the name of the user with the given uid.
 
     If uid is None, return the name of the current user."""
@@ -604,7 +658,7 b' def username(uid=None):'
         return b'%d' % uid
 
 
-def groupname(gid=None):
+def groupname(gid: Optional[int] = None) -> Optional[bytes]:
     """Return the name of the group with the given gid.
 
     If gid is None, return the name of the current group."""
@@ -617,7 +671,7 b' def groupname(gid=None):'
         return pycompat.bytestr(gid)
 
 
-def groupmembers(name):
+def groupmembers(name: bytes) -> List[bytes]:
     """Return the list of members of the group with the given
     name, KeyError if the group does not exist.
     """
@@ -625,23 +679,27 b' def groupmembers(name):'
     return pycompat.rapply(pycompat.fsencode, list(grp.getgrnam(name).gr_mem))
 
 
-def spawndetached(args):
+def spawndetached(args: List[bytes]) -> int:
     return os.spawnvp(os.P_NOWAIT | getattr(os, 'P_DETACH', 0), args[0], args)
 
 
-def gethgcmd():
+def gethgcmd():  # TODO: convert to bytes, like on Windows?
     return sys.argv[:1]
 
 
-def makedir(path, notindexed):
+def makedir(path: bytes, notindexed: bool) -> None:
     os.mkdir(path)
 
 
-def lookupreg(key, name=None, scope=None):
+def lookupreg(
+    key: bytes,
+    name: Optional[bytes] = None,
+    scope: Optional[Union[int, Iterable[int]]] = None,
+) -> Optional[bytes]:
     return None
 
 
-def hidewindow():
+def hidewindow() -> None:
     """Hide current shell window.
 
     Used to hide the window opened when starting asynchronous
@@ -651,15 +709,15 b' def hidewindow():'
 
 
 class cachestat:
-    def __init__(self, path):
+    def __init__(self, path: bytes) -> None:
         self.stat = os.stat(path)
 
-    def cacheable(self):
+    def cacheable(self) -> bool:
         return bool(self.stat.st_ino)
 
     __hash__ = object.__hash__
 
-    def __eq__(self, other):
+    def __eq__(self, other: Any) -> bool:
         try:
             # Only dev, ino, size, mtime and atime are likely to change. Out
             # of these, we shouldn't compare atime but should compare the
@@ -680,18 +738,18 b' class cachestat:'
         except AttributeError:
             return False
 
-    def __ne__(self, other):
+    def __ne__(self, other: Any) -> bool:
         return not self == other
 
 
-def statislink(st):
+def statislink(st: Optional[os.stat_result]) -> bool:
     '''check whether a stat result is a symlink'''
-    return st and stat.S_ISLNK(st.st_mode)
+    return stat.S_ISLNK(st.st_mode) if st else False
 
 
-def statisexec(st):
+def statisexec(st: Optional[os.stat_result]) -> bool:
     '''check whether a stat result is an executable file'''
-    return st and (st.st_mode & 0o100 != 0)
+    return (st.st_mode & 0o100 != 0) if st else False
 
 
 def poll(fds):
@@ -708,7 +766,7 b' def poll(fds):'
     return sorted(list(set(sum(res, []))))
 
 
-def readpipe(pipe):
+def readpipe(pipe) -> bytes:
     """Read all available data from a pipe."""
     # We can't fstat() a pipe because Linux will always report 0.
     # So, we set the pipe to non-blocking mode and read everything
@@ -733,7 +791,7 b' def readpipe(pipe):'
         fcntl.fcntl(pipe, fcntl.F_SETFL, oldflags)
 
 
-def bindunixsocket(sock, path):
+def bindunixsocket(sock, path: bytes) -> None:
     """Bind the UNIX domain socket to the specified path"""
     # use relative path instead of full path at bind() if possible, since
     # AF_UNIX path has very small length limit (107 chars) on common

mercurial/pure/bdiff.py

@@ -10,8 +10,13 b' import difflib'
 import re
 import struct
 
+from typing import (
+    List,
+    Tuple,
+)
 
-def splitnewlines(text):
+
+def splitnewlines(text: bytes) -> List[bytes]:
     '''like str.splitlines, but only split on newlines.'''
     lines = [l + b'\n' for l in text.split(b'\n')]
     if lines:
@@ -22,7 +27,9 b' def splitnewlines(text):'
     return lines
 
 
-def _normalizeblocks(a, b, blocks):
+def _normalizeblocks(
+    a: List[bytes], b: List[bytes], blocks
+) -> List[Tuple[int, int, int]]:
     prev = None
     r = []
     for curr in blocks:
@@ -57,7 +64,7 b' def _normalizeblocks(a, b, blocks):'
     return r
 
 
-def bdiff(a, b):
+def bdiff(a: bytes, b: bytes) -> bytes:
     a = bytes(a).splitlines(True)
     b = bytes(b).splitlines(True)
 
@@ -84,7 +91,7 b' def bdiff(a, b):'
     return b"".join(bin)
 
 
-def blocks(a, b):
+def blocks(a: bytes, b: bytes) -> List[Tuple[int, int, int, int]]:
     an = splitnewlines(a)
     bn = splitnewlines(b)
     d = difflib.SequenceMatcher(None, an, bn).get_matching_blocks()
@@ -92,7 +99,7 b' def blocks(a, b):'
     return [(i, i + n, j, j + n) for (i, j, n) in d]
 
 
-def fixws(text, allws):
+def fixws(text: bytes, allws: bool) -> bytes:
     if allws:
         text = re.sub(b'[ \t\r]+', b'', text)
     else:

mercurial/pure/mpatch.py

@@ -9,6 +9,11 b''
 import io
 import struct
 
+from typing import (
+    List,
+    Tuple,
+)
+
 
 stringio = io.BytesIO
 
@@ -28,7 +33,9 b' class mpatchError(Exception):'
 # temporary string buffers.
 
 
-def _pull(dst, src, l):  # pull l bytes from src
+def _pull(
+    dst: List[Tuple[int, int]], src: List[Tuple[int, int]], l: int
+) -> None:  # pull l bytes from src
     while l:
         f = src.pop()
         if f[0] > l:  # do we need to split?
@@ -39,7 +46,7 b' def _pull(dst, src, l): # pull l bytes '
         l -= f[0]
 
 
-def _move(m, dest, src, count):
+def _move(m: stringio, dest: int, src: int, count: int) -> None:
     """move count bytes from src to dest
 
     The file pointer is left at the end of dest.
@@ -50,7 +57,9 b' def _move(m, dest, src, count):'
     m.write(buf)
 
 
-def _collect(m, buf, list):
+def _collect(
+    m: stringio, buf: int, list: List[Tuple[int, int]]
+) -> Tuple[int, int]:
     start = buf
     for l, p in reversed(list):
         _move(m, buf, p, l)
@@ -58,7 +67,7 b' def _collect(m, buf, list):'
     return (buf - start, start)
 
 
-def patches(a, bins):
+def patches(a: bytes, bins: List[bytes]) -> bytes:
     if not bins:
         return a
 
@@ -111,7 +120,7 b' def patches(a, bins):'
     return m.read(t[0])
 
 
-def patchedsize(orig, delta):
+def patchedsize(orig: int, delta: bytes) -> int:
     outlen, last, bin = 0, 0, 0
     binend = len(delta)
     data = 12

mercurial/pure/parsers.py

@@ -435,6 +435,11 b' class DirstateItem:'
         return self._wc_tracked and not (self._p1_tracked or self._p2_info)
 
     @property
+    def modified(self):
+        """True if the file has been modified"""
+        return self._wc_tracked and self._p1_tracked and self._p2_info
+
+    @property
     def maybe_clean(self):
         """True if the file has a chance to be in the "clean" state"""
         if not self._wc_tracked:

mercurial/pycompat.py

@@ -28,6 +28,24 b' import sys'
 import tempfile
 import xmlrpc.client as xmlrpclib
 
+from typing import (
+    Any,
+    AnyStr,
+    BinaryIO,
+    Dict,
+    Iterable,
+    Iterator,
+    List,
+    Mapping,
+    NoReturn,
+    Optional,
+    Sequence,
+    Tuple,
+    Type,
+    TypeVar,
+    cast,
+    overload,
+)
 
 ispy3 = sys.version_info[0] >= 3
 ispypy = '__pypy__' in sys.builtin_module_names
@@ -38,6 +56,10 b' if not globals(): # hide this from non-'
 
     TYPE_CHECKING = typing.TYPE_CHECKING
 
+_GetOptResult = Tuple[List[Tuple[bytes, bytes]], List[bytes]]
+_T0 = TypeVar('_T0')
+_Tbytestr = TypeVar('_Tbytestr', bound='bytestr')
+
 
 def future_set_exception_info(f, exc_info):
     f.set_exception(exc_info[0])
@@ -46,7 +68,7 b' def future_set_exception_info(f, exc_inf'
 FileNotFoundError = builtins.FileNotFoundError
 
 
-def identity(a):
+def identity(a: _T0) -> _T0:
     return a
 
 
@@ -94,21 +116,17 b" if os.name == r'nt':"
 
 fsencode = os.fsencode
 fsdecode = os.fsdecode
-oscurdir = os.curdir.encode('ascii')
-oslinesep = os.linesep.encode('ascii')
-osname = os.name.encode('ascii')
-ospathsep = os.pathsep.encode('ascii')
-ospardir = os.pardir.encode('ascii')
-ossep = os.sep.encode('ascii')
-osaltsep = os.altsep
-if osaltsep:
-    osaltsep = osaltsep.encode('ascii')
-osdevnull = os.devnull.encode('ascii')
+oscurdir: bytes = os.curdir.encode('ascii')
+oslinesep: bytes = os.linesep.encode('ascii')
+osname: bytes = os.name.encode('ascii')
+ospathsep: bytes = os.pathsep.encode('ascii')
+ospardir: bytes = os.pardir.encode('ascii')
+ossep: bytes = os.sep.encode('ascii')
+osaltsep: Optional[bytes] = os.altsep.encode('ascii') if os.altsep else None
+osdevnull: bytes = os.devnull.encode('ascii')
 
-sysplatform = sys.platform.encode('ascii')
-sysexecutable = sys.executable
-if sysexecutable:
-    sysexecutable = os.fsencode(sysexecutable)
+sysplatform: bytes = sys.platform.encode('ascii')
+sysexecutable: bytes = os.fsencode(sys.executable) if sys.executable else b''
 
 
 def maplist(*args):
@@ -128,7 +146,7 b' getargspec = inspect.getfullargspec'
 
 long = int
 
-if getattr(sys, 'argv', None) is not None:
+if builtins.getattr(sys, 'argv', None) is not None:
     # On POSIX, the char** argv array is converted to Python str using
     # Py_DecodeLocale(). The inverse of this is Py_EncodeLocale(), which
     # isn't directly callable from Python code. In practice, os.fsencode()
@@ -143,6 +161,7 b" if getattr(sys, 'argv', None) is not Non"
     # (this is how Python 2 worked). To get that, we encode with the mbcs
     # encoding, which will pass CP_ACP to the underlying Windows API to
     # produce bytes.
+    sysargv: List[bytes] = []
     if os.name == r'nt':
         sysargv = [a.encode("mbcs", "ignore") for a in sys.argv]
     else:
@@ -211,38 +230,53 b' class bytestr(bytes):'
     # https://github.com/google/pytype/issues/500
     if TYPE_CHECKING:
 
-        def __init__(self, s=b''):
+        def __init__(self, s: object = b'') -> None:
             pass
 
-    def __new__(cls, s=b''):
+    def __new__(cls: Type[_Tbytestr], s: object = b'') -> _Tbytestr:
         if isinstance(s, bytestr):
             return s
         if not isinstance(
             s, (bytes, bytearray)
-        ) and not hasattr(  # hasattr-py3-only
+        ) and not builtins.hasattr(  # hasattr-py3-only
             s, u'__bytes__'
         ):
             s = str(s).encode('ascii')
         return bytes.__new__(cls, s)
 
-    def __getitem__(self, key):
+    # The base class uses `int` return in py3, but the point of this class is to
+    # behave like py2.
+    def __getitem__(self, key) -> bytes:  # pytype: disable=signature-mismatch
         s = bytes.__getitem__(self, key)
         if not isinstance(s, bytes):
             s = bytechr(s)
         return s
 
-    def __iter__(self):
+    # The base class expects `Iterator[int]` return in py3, but the point of
+    # this class is to behave like py2.
+    def __iter__(self) -> Iterator[bytes]:  # pytype: disable=signature-mismatch
         return iterbytestr(bytes.__iter__(self))
 
-    def __repr__(self):
+    def __repr__(self) -> str:
         return bytes.__repr__(self)[1:]  # drop b''
 
 
-def iterbytestr(s):
+def iterbytestr(s: Iterable[int]) -> Iterator[bytes]:
     """Iterate bytes as if it were a str object of Python 2"""
     return map(bytechr, s)
 
 
+if TYPE_CHECKING:
+
+    @overload
+    def maybebytestr(s: bytes) -> bytestr:
+        ...
+
+    @overload
+    def maybebytestr(s: _T0) -> _T0:
+        ...
+
+
 def maybebytestr(s):
     """Promote bytes to bytestr"""
     if isinstance(s, bytes):
@@ -250,7 +284,7 b' def maybebytestr(s):'
     return s
 
 
-def sysbytes(s):
+def sysbytes(s: AnyStr) -> bytes:
     """Convert an internal str (e.g. keyword, __doc__) back to bytes
 
     This never raises UnicodeEncodeError, but only ASCII characters
@@ -261,7 +295,7 b' def sysbytes(s):'
     return s.encode('utf-8')
 
 
-def sysstr(s):
+def sysstr(s: AnyStr) -> str:
     """Return a keyword str to be passed to Python functions such as
     getattr() and str.encode()
 
@@ -274,29 +308,29 b' def sysstr(s):'
     return s.decode('latin-1')
 
 
-def strurl(url):
+def strurl(url: AnyStr) -> str:
     """Converts a bytes url back to str"""
     if isinstance(url, bytes):
         return url.decode('ascii')
     return url
 
 
-def bytesurl(url):
+def bytesurl(url: AnyStr) -> bytes:
     """Converts a str url to bytes by encoding in ascii"""
     if isinstance(url, str):
         return url.encode('ascii')
     return url
 
 
-def raisewithtb(exc, tb):
+def raisewithtb(exc: BaseException, tb) -> NoReturn:
     """Raise exception with the given traceback"""
     raise exc.with_traceback(tb)
 
 
-def getdoc(obj):
+def getdoc(obj: object) -> Optional[bytes]:
     """Get docstring as bytes; may be None so gettext() won't confuse it
     with _('')"""
-    doc = getattr(obj, '__doc__', None)
+    doc = builtins.getattr(obj, '__doc__', None)
     if doc is None:
         return doc
     return sysbytes(doc)
@@ -319,14 +353,22 b' xrange = builtins.range'
 unicode = str
 
 
-def open(name, mode=b'r', buffering=-1, encoding=None):
+def open(
+    name,
+    mode: AnyStr = b'r',
+    buffering: int = -1,
+    encoding: Optional[str] = None,
+) -> Any:
+    # TODO: assert binary mode, and cast result to BinaryIO?
     return builtins.open(name, sysstr(mode), buffering, encoding)
 
 
 safehasattr = _wrapattrfunc(builtins.hasattr)
 
 
-def _getoptbwrapper(orig, args, shortlist, namelist):
+def _getoptbwrapper(
+    orig, args: Sequence[bytes], shortlist: bytes, namelist: Sequence[bytes]
+) -> _GetOptResult:
     """
     Takes bytes arguments, converts them to unicode, pass them to
     getopt.getopt(), convert the returned values back to bytes and then
@@ -342,7 +384,7 b' def _getoptbwrapper(orig, args, shortlis'
     return opts, args
 
 
-def strkwargs(dic):
+def strkwargs(dic: Mapping[bytes, _T0]) -> Dict[str, _T0]:
     """
     Converts the keys of a python dictonary to str i.e. unicodes so that
     they can be passed as keyword arguments as dictionaries with bytes keys
@@ -352,7 +394,7 b' def strkwargs(dic):'
     return dic
 
 
-def byteskwargs(dic):
+def byteskwargs(dic: Mapping[str, _T0]) -> Dict[bytes, _T0]:
     """
     Converts keys of python dictionaries to bytes as they were converted to
     str to pass that dictonary as a keyword argument on Python 3.
@@ -362,7 +404,9 b' def byteskwargs(dic):'
 
 
 # TODO: handle shlex.shlex().
-def shlexsplit(s, comments=False, posix=True):
+def shlexsplit(
+    s: bytes, comments: bool = False, posix: bool = True
+) -> List[bytes]:
     """
     Takes bytes argument, convert it to str i.e. unicodes, pass that into
     shlex.split(), convert the returned value to bytes and return that for
@@ -377,46 +421,59 b' itervalues = lambda x: x.values()'
 
 json_loads = json.loads
 
-isjython = sysplatform.startswith(b'java')
+isjython: bool = sysplatform.startswith(b'java')
 
-isdarwin = sysplatform.startswith(b'darwin')
-islinux = sysplatform.startswith(b'linux')
-isposix = osname == b'posix'
-iswindows = osname == b'nt'
+isdarwin: bool = sysplatform.startswith(b'darwin')
+islinux: bool = sysplatform.startswith(b'linux')
+isposix: bool = osname == b'posix'
+iswindows: bool = osname == b'nt'
 
 
-def getoptb(args, shortlist, namelist):
+def getoptb(
+    args: Sequence[bytes], shortlist: bytes, namelist: Sequence[bytes]
+) -> _GetOptResult:
     return _getoptbwrapper(getopt.getopt, args, shortlist, namelist)
 
 
-def gnugetoptb(args, shortlist, namelist):
+def gnugetoptb(
+    args: Sequence[bytes], shortlist: bytes, namelist: Sequence[bytes]
+) -> _GetOptResult:
     return _getoptbwrapper(getopt.gnu_getopt, args, shortlist, namelist)
 
 
-def mkdtemp(suffix=b'', prefix=b'tmp', dir=None):
+def mkdtemp(
+    suffix: bytes = b'', prefix: bytes = b'tmp', dir: Optional[bytes] = None
+) -> bytes:
     return tempfile.mkdtemp(suffix, prefix, dir)
 
 
 # text=True is not supported; use util.from/tonativeeol() instead
-def mkstemp(suffix=b'', prefix=b'tmp', dir=None):
+def mkstemp(
+    suffix: bytes = b'', prefix: bytes = b'tmp', dir: Optional[bytes] = None
+) -> Tuple[int, bytes]:
     return tempfile.mkstemp(suffix, prefix, dir)
 
 
 # TemporaryFile does not support an "encoding=" argument on python2.
 # This wrapper file are always open in byte mode.
-def unnamedtempfile(mode=None, *args, **kwargs):
+def unnamedtempfile(mode: Optional[bytes] = None, *args, **kwargs) -> BinaryIO:
     if mode is None:
         mode = 'w+b'
     else:
         mode = sysstr(mode)
     assert 'b' in mode
-    return tempfile.TemporaryFile(mode, *args, **kwargs)
+    return cast(BinaryIO, tempfile.TemporaryFile(mode, *args, **kwargs))
 
 
 # NamedTemporaryFile does not support an "encoding=" argument on python2.
 # This wrapper file are always open in byte mode.
 def namedtempfile(
-    mode=b'w+b', bufsize=-1, suffix=b'', prefix=b'tmp', dir=None, delete=True
+    mode: bytes = b'w+b',
+    bufsize: int = -1,
+    suffix: bytes = b'',
+    prefix: bytes = b'tmp',
+    dir: Optional[bytes] = None,
+    delete: bool = True,
 ):
     mode = sysstr(mode)
     assert 'b' in mode

mercurial/revlog.py

@@ -38,12 +38,15 b' from .revlogutils.constants import ('
     COMP_MODE_DEFAULT,
     COMP_MODE_INLINE,
     COMP_MODE_PLAIN,
+    DELTA_BASE_REUSE_NO,
+    DELTA_BASE_REUSE_TRY,
     ENTRY_RANK,
     FEATURES_BY_VERSION,
     FLAG_GENERALDELTA,
     FLAG_INLINE_DATA,
     INDEX_HEADER,
     KIND_CHANGELOG,
+    KIND_FILELOG,
     RANK_UNKNOWN,
     REVLOGV0,
     REVLOGV1,
@@ -125,7 +128,7 b" rustrevlog = policy.importrust('revlog')"
 # Aliased for performance.
 _zlibdecompress = zlib.decompress
 
-# max size of revlog with inline data
+# max size of inline data embedded into a revlog
 _maxinline = 131072
 
 # Flag processors for REVIDX_ELLIPSIS.
@@ -347,6 +350,7 b' class revlog:'
         self._chunkcachesize = 65536
         self._maxchainlen = None
         self._deltabothparents = True
+        self._candidate_group_chunk_size = 0
         self._debug_delta = False
         self.index = None
         self._docket = None
@@ -363,6 +367,11 b' class revlog:'
         self._srdensitythreshold = 0.50
         self._srmingapsize = 262144
 
+        # other optionnals features
+
+        # might remove rank configuration once the computation has no impact
+        self._compute_rank = False
+
         # Make copy of flag processors so each revlog instance can support
         # custom flags.
         self._flagprocessors = dict(flagutil.flagprocessors)
@@ -404,6 +413,7 b' class revlog:'
 
         if b'changelogv2' in opts and self.revlog_kind == KIND_CHANGELOG:
             new_header = CHANGELOGV2
+            self._compute_rank = opts.get(b'changelogv2.compute-rank', True)
         elif b'revlogv2' in opts:
             new_header = REVLOGV2
         elif b'revlogv1' in opts:
@@ -421,6 +431,9 b' class revlog:'
             self._maxchainlen = opts[b'maxchainlen']
         if b'deltabothparents' in opts:
             self._deltabothparents = opts[b'deltabothparents']
+        dps_cgds = opts.get(b'delta-parent-search.candidate-group-chunk-size')
+        if dps_cgds:
+            self._candidate_group_chunk_size = dps_cgds
         self._lazydelta = bool(opts.get(b'lazydelta', True))
         self._lazydeltabase = False
         if self._lazydelta:
@@ -505,7 +518,6 b' class revlog:'
             self._docket = docket
             self._docket_file = entry_point
         else:
-            entry_data = b''
             self._initempty = True
             entry_data = self._get_data(entry_point, mmapindexthreshold)
             if len(entry_data) > 0:
@@ -653,9 +665,12 b' class revlog:'
     @util.propertycache
     def display_id(self):
         """The public facing "ID" of the revlog that we use in message"""
-        # Maybe we should build a user facing representation of
-        # revlog.target instead of using `self.radix`
-        return self.radix
+        if self.revlog_kind == KIND_FILELOG:
+            # Reference the file without the "data/" prefix, so it is familiar
+            # to the user.
+            return self.target[1]
+        else:
+            return self.radix
 
     def _get_decompressor(self, t):
         try:
@@ -2445,6 +2460,16 b' class revlog:'
                 self, write_debug=write_debug
             )
 
+        if cachedelta is not None and len(cachedelta) == 2:
+            # If the cached delta has no information about how it should be
+            # reused, add the default reuse instruction according to the
+            # revlog's configuration.
+            if self._generaldelta and self._lazydeltabase:
+                delta_base_reuse = DELTA_BASE_REUSE_TRY
+            else:
+                delta_base_reuse = DELTA_BASE_REUSE_NO
+            cachedelta = (cachedelta[0], cachedelta[1], delta_base_reuse)
+
         revinfo = revlogutils.revisioninfo(
             node,
             p1,
@@ -2492,7 +2517,7 b' class revlog:'
             sidedata_offset = 0
 
         rank = RANK_UNKNOWN
-        if self._format_version == CHANGELOGV2:
+        if self._compute_rank:
             if (p1r, p2r) == (nullrev, nullrev):
                 rank = 1
             elif p1r != nullrev and p2r == nullrev:
@@ -2637,6 +2662,8 b' class revlog:'
         alwayscache=False,
         addrevisioncb=None,
         duplicaterevisioncb=None,
+        debug_info=None,
+        delta_base_reuse_policy=None,
     ):
         """
         add a delta group
@@ -2652,6 +2679,14 b' class revlog:'
         if self._adding_group:
             raise error.ProgrammingError(b'cannot nest addgroup() calls')
 
+        # read the default delta-base reuse policy from revlog config if the
+        # group did not specify one.
+        if delta_base_reuse_policy is None:
+            if self._generaldelta and self._lazydeltabase:
+                delta_base_reuse_policy = DELTA_BASE_REUSE_TRY
+            else:
+                delta_base_reuse_policy = DELTA_BASE_REUSE_NO
+
         self._adding_group = True
         empty = True
         try:
@@ -2662,6 +2697,7 b' class revlog:'
                 deltacomputer = deltautil.deltacomputer(
                     self,
                     write_debug=write_debug,
+                    debug_info=debug_info,
                 )
                 # loop through our set of deltas
                 for data in deltas:
@@ -2731,7 +2767,7 b' class revlog:'
                         p1,
                         p2,
                         flags,
-                        (baserev, delta),
+                        (baserev, delta, delta_base_reuse_policy),
                         alwayscache=alwayscache,
                         deltacomputer=deltacomputer,
                         sidedata=sidedata,
@@ -2886,6 +2922,7 b' class revlog:'
         assumehaveparentrevisions=False,
         deltamode=repository.CG_DELTAMODE_STD,
         sidedata_helpers=None,
+        debug_info=None,
     ):
         if nodesorder not in (b'nodes', b'storage', b'linear', None):
             raise error.ProgrammingError(
@@ -2915,6 +2952,7 b' class revlog:'
             revisiondata=revisiondata,
             assumehaveparentrevisions=assumehaveparentrevisions,
             sidedata_helpers=sidedata_helpers,
+            debug_info=debug_info,
         )
 
     DELTAREUSEALWAYS = b'always'

mercurial/revlogutils/__init__.py

@@ -67,7 +67,7 b' class revisioninfo:'
     node:       expected hash of the revision
     p1, p2:     parent revs of the revision
     btext:      built text cache consisting of a one-element list
-    cachedelta: (baserev, uncompressed_delta) or None
+    cachedelta: (baserev, uncompressed_delta, usage_mode) or None
     flags:      flags associated to the revision storage
 
     One of btext[0] or cachedelta must be set.

mercurial/revlogutils/constants.py

@@ -301,3 +301,18 b' FEATURES_BY_VERSION = {'
 
 
 SPARSE_REVLOG_MAX_CHAIN_LENGTH = 1000
+
+### What should be done with a cached delta and its base ?
+
+# Ignore the cache when considering candidates.
+#
+# The cached delta might be used, but the delta base will not be scheduled for
+# usage earlier than in "normal" order.
+DELTA_BASE_REUSE_NO = 0
+
+# Prioritize trying the cached delta base
+#
+# The delta base will be tested for validy first. So that the cached deltas get
+# used when possible.
+DELTA_BASE_REUSE_TRY = 1
+DELTA_BASE_REUSE_FORCE = 2

mercurial/revlogutils/debug.py

@@ -6,12 +6,19 b''
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
+import collections
+import string
+
 from .. import (
+    mdiff,
     node as nodemod,
+    revlogutils,
+    util,
 )
 
 from . import (
     constants,
+    deltas as deltautil,
 )
 
 INDEX_ENTRY_DEBUG_COLUMN = []
@@ -216,3 +223,499 b' def debug_index('
         fm.plain(b'\n')
 
     fm.end()
+
+
+def dump(ui, revlog):
+    """perform the work for `hg debugrevlog --dump"""
+    # XXX seems redundant with debug index ?
+    r = revlog
+    numrevs = len(r)
+    ui.write(
+        (
+            b"# rev p1rev p2rev start   end deltastart base   p1   p2"
+            b" rawsize totalsize compression heads chainlen\n"
+        )
+    )
+    ts = 0
+    heads = set()
+
+    for rev in range(numrevs):
+        dbase = r.deltaparent(rev)
+        if dbase == -1:
+            dbase = rev
+        cbase = r.chainbase(rev)
+        clen = r.chainlen(rev)
+        p1, p2 = r.parentrevs(rev)
+        rs = r.rawsize(rev)
+        ts = ts + rs
+        heads -= set(r.parentrevs(rev))
+        heads.add(rev)
+        try:
+            compression = ts / r.end(rev)
+        except ZeroDivisionError:
+            compression = 0
+        ui.write(
+            b"%5d %5d %5d %5d %5d %10d %4d %4d %4d %7d %9d "
+            b"%11d %5d %8d\n"
+            % (
+                rev,
+                p1,
+                p2,
+                r.start(rev),
+                r.end(rev),
+                r.start(dbase),
+                r.start(cbase),
+                r.start(p1),
+                r.start(p2),
+                rs,
+                ts,
+                compression,
+                len(heads),
+                clen,
+            )
+        )
+
+
+def debug_revlog(ui, revlog):
+    """code for `hg debugrevlog`"""
+    r = revlog
+    format = r._format_version
+    v = r._format_flags
+    flags = []
+    gdelta = False
+    if v & constants.FLAG_INLINE_DATA:
+        flags.append(b'inline')
+    if v & constants.FLAG_GENERALDELTA:
+        gdelta = True
+        flags.append(b'generaldelta')
+    if not flags:
+        flags = [b'(none)']
+
+    ### the total size of stored content if incompressed.
+    full_text_total_size = 0
+    ### tracks merge vs single parent
+    nummerges = 0
+
+    ### tracks ways the "delta" are build
+    # nodelta
+    numempty = 0
+    numemptytext = 0
+    numemptydelta = 0
+    # full file content
+    numfull = 0
+    # intermediate snapshot against a prior snapshot
+    numsemi = 0
+    # snapshot count per depth
+    numsnapdepth = collections.defaultdict(lambda: 0)
+    # number of snapshots with a non-ancestor delta
+    numsnapdepth_nad = collections.defaultdict(lambda: 0)
+    # delta against previous revision
+    numprev = 0
+    # delta against prev, where prev is a non-ancestor
+    numprev_nad = 0
+    # delta against first or second parent (not prev)
+    nump1 = 0
+    nump2 = 0
+    # delta against neither prev nor parents
+    numother = 0
+    # delta against other that is a non-ancestor
+    numother_nad = 0
+    # delta against prev that are also first or second parent
+    # (details of `numprev`)
+    nump1prev = 0
+    nump2prev = 0
+
+    # data about delta chain of each revs
+    chainlengths = []
+    chainbases = []
+    chainspans = []
+
+    # data about each revision
+    datasize = [None, 0, 0]
+    fullsize = [None, 0, 0]
+    semisize = [None, 0, 0]
+    # snapshot count per depth
+    snapsizedepth = collections.defaultdict(lambda: [None, 0, 0])
+    deltasize = [None, 0, 0]
+    chunktypecounts = {}
+    chunktypesizes = {}
+
+    def addsize(size, l):
+        if l[0] is None or size < l[0]:
+            l[0] = size
+        if size > l[1]:
+            l[1] = size
+        l[2] += size
+
+    numrevs = len(r)
+    for rev in range(numrevs):
+        p1, p2 = r.parentrevs(rev)
+        delta = r.deltaparent(rev)
+        if format > 0:
+            s = r.rawsize(rev)
+            full_text_total_size += s
+            addsize(s, datasize)
+        if p2 != nodemod.nullrev:
+            nummerges += 1
+        size = r.length(rev)
+        if delta == nodemod.nullrev:
+            chainlengths.append(0)
+            chainbases.append(r.start(rev))
+            chainspans.append(size)
+            if size == 0:
+                numempty += 1
+                numemptytext += 1
+            else:
+                numfull += 1
+                numsnapdepth[0] += 1
+                addsize(size, fullsize)
+                addsize(size, snapsizedepth[0])
+        else:
+            nad = (
+                delta != p1 and delta != p2 and not r.isancestorrev(delta, rev)
+            )
+            chainlengths.append(chainlengths[delta] + 1)
+            baseaddr = chainbases[delta]
+            revaddr = r.start(rev)
+            chainbases.append(baseaddr)
+            chainspans.append((revaddr - baseaddr) + size)
+            if size == 0:
+                numempty += 1
+                numemptydelta += 1
+            elif r.issnapshot(rev):
+                addsize(size, semisize)
+                numsemi += 1
+                depth = r.snapshotdepth(rev)
+                numsnapdepth[depth] += 1
+                if nad:
+                    numsnapdepth_nad[depth] += 1
+                addsize(size, snapsizedepth[depth])
+            else:
+                addsize(size, deltasize)
+                if delta == rev - 1:
+                    numprev += 1
+                    if delta == p1:
+                        nump1prev += 1
+                    elif delta == p2:
+                        nump2prev += 1
+                    elif nad:
+                        numprev_nad += 1
+                elif delta == p1:
+                    nump1 += 1
+                elif delta == p2:
+                    nump2 += 1
+                elif delta != nodemod.nullrev:
+                    numother += 1
+                    numother_nad += 1
+
+        # Obtain data on the raw chunks in the revlog.
+        if util.safehasattr(r, '_getsegmentforrevs'):
+            segment = r._getsegmentforrevs(rev, rev)[1]
+        else:
+            segment = r._revlog._getsegmentforrevs(rev, rev)[1]
+        if segment:
+            chunktype = bytes(segment[0:1])
+        else:
+            chunktype = b'empty'
+
+        if chunktype not in chunktypecounts:
+            chunktypecounts[chunktype] = 0
+            chunktypesizes[chunktype] = 0
+
+        chunktypecounts[chunktype] += 1
+        chunktypesizes[chunktype] += size
+
+    # Adjust size min value for empty cases
+    for size in (datasize, fullsize, semisize, deltasize):
+        if size[0] is None:
+            size[0] = 0
+
+    numdeltas = numrevs - numfull - numempty - numsemi
+    numoprev = numprev - nump1prev - nump2prev - numprev_nad
+    num_other_ancestors = numother - numother_nad
+    totalrawsize = datasize[2]
+    datasize[2] /= numrevs
+    fulltotal = fullsize[2]
+    if numfull == 0:
+        fullsize[2] = 0
+    else:
+        fullsize[2] /= numfull
+    semitotal = semisize[2]
+    snaptotal = {}
+    if numsemi > 0:
+        semisize[2] /= numsemi
+    for depth in snapsizedepth:
+        snaptotal[depth] = snapsizedepth[depth][2]
+        snapsizedepth[depth][2] /= numsnapdepth[depth]
+
+    deltatotal = deltasize[2]
+    if numdeltas > 0:
+        deltasize[2] /= numdeltas
+    totalsize = fulltotal + semitotal + deltatotal
+    avgchainlen = sum(chainlengths) / numrevs
+    maxchainlen = max(chainlengths)
+    maxchainspan = max(chainspans)
+    compratio = 1
+    if totalsize:
+        compratio = totalrawsize / totalsize
+
+    basedfmtstr = b'%%%dd\n'
+    basepcfmtstr = b'%%%dd %s(%%5.2f%%%%)\n'
+
+    def dfmtstr(max):
+        return basedfmtstr % len(str(max))
+
+    def pcfmtstr(max, padding=0):
+        return basepcfmtstr % (len(str(max)), b' ' * padding)
+
+    def pcfmt(value, total):
+        if total:
+            return (value, 100 * float(value) / total)
+        else:
+            return value, 100.0
+
+    ui.writenoi18n(b'format : %d\n' % format)
+    ui.writenoi18n(b'flags  : %s\n' % b', '.join(flags))
+
+    ui.write(b'\n')
+    fmt = pcfmtstr(totalsize)
+    fmt2 = dfmtstr(totalsize)
+    ui.writenoi18n(b'revisions     : ' + fmt2 % numrevs)
+    ui.writenoi18n(b'    merges    : ' + fmt % pcfmt(nummerges, numrevs))
+    ui.writenoi18n(
+        b'    normal    : ' + fmt % pcfmt(numrevs - nummerges, numrevs)
+    )
+    ui.writenoi18n(b'revisions     : ' + fmt2 % numrevs)
+    ui.writenoi18n(b'    empty     : ' + fmt % pcfmt(numempty, numrevs))
+    ui.writenoi18n(
+        b'                   text  : '
+        + fmt % pcfmt(numemptytext, numemptytext + numemptydelta)
+    )
+    ui.writenoi18n(
+        b'                   delta : '
+        + fmt % pcfmt(numemptydelta, numemptytext + numemptydelta)
+    )
+    ui.writenoi18n(
+        b'    snapshot  : ' + fmt % pcfmt(numfull + numsemi, numrevs)
+    )
+    for depth in sorted(numsnapdepth):
+        base = b'      lvl-%-3d :       ' % depth
+        count = fmt % pcfmt(numsnapdepth[depth], numrevs)
+        pieces = [base, count]
+        if numsnapdepth_nad[depth]:
+            pieces[-1] = count = count[:-1]  # drop the final '\n'
+            more = b'  non-ancestor-bases: '
+            anc_count = fmt
+            anc_count %= pcfmt(numsnapdepth_nad[depth], numsnapdepth[depth])
+            pieces.append(more)
+            pieces.append(anc_count)
+        ui.write(b''.join(pieces))
+    ui.writenoi18n(b'    deltas    : ' + fmt % pcfmt(numdeltas, numrevs))
+    ui.writenoi18n(b'revision size : ' + fmt2 % totalsize)
+    ui.writenoi18n(
+        b'    snapshot  : ' + fmt % pcfmt(fulltotal + semitotal, totalsize)
+    )
+    for depth in sorted(numsnapdepth):
+        ui.write(
+            (b'      lvl-%-3d :       ' % depth)
+            + fmt % pcfmt(snaptotal[depth], totalsize)
+        )
+    ui.writenoi18n(b'    deltas    : ' + fmt % pcfmt(deltatotal, totalsize))
+
+    letters = string.ascii_letters.encode('ascii')
+
+    def fmtchunktype(chunktype):
+        if chunktype == b'empty':
+            return b'    %s     : ' % chunktype
+        elif chunktype in letters:
+            return b'    0x%s (%s)  : ' % (nodemod.hex(chunktype), chunktype)
+        else:
+            return b'    0x%s      : ' % nodemod.hex(chunktype)
+
+    ui.write(b'\n')
+    ui.writenoi18n(b'chunks        : ' + fmt2 % numrevs)
+    for chunktype in sorted(chunktypecounts):
+        ui.write(fmtchunktype(chunktype))
+        ui.write(fmt % pcfmt(chunktypecounts[chunktype], numrevs))
+    ui.writenoi18n(b'chunks size   : ' + fmt2 % totalsize)
+    for chunktype in sorted(chunktypecounts):
+        ui.write(fmtchunktype(chunktype))
+        ui.write(fmt % pcfmt(chunktypesizes[chunktype], totalsize))
+
+    ui.write(b'\n')
+    b_total = b"%d" % full_text_total_size
+    p_total = []
+    while len(b_total) > 3:
+        p_total.append(b_total[-3:])
+        b_total = b_total[:-3]
+    p_total.append(b_total)
+    p_total.reverse()
+    b_total = b' '.join(p_total)
+
+    ui.write(b'\n')
+    ui.writenoi18n(b'total-stored-content: %s bytes\n' % b_total)
+    ui.write(b'\n')
+    fmt = dfmtstr(max(avgchainlen, maxchainlen, maxchainspan, compratio))
+    ui.writenoi18n(b'avg chain length  : ' + fmt % avgchainlen)
+    ui.writenoi18n(b'max chain length  : ' + fmt % maxchainlen)
+    ui.writenoi18n(b'max chain reach   : ' + fmt % maxchainspan)
+    ui.writenoi18n(b'compression ratio : ' + fmt % compratio)
+
+    if format > 0:
+        ui.write(b'\n')
+        ui.writenoi18n(
+            b'uncompressed data size (min/max/avg) : %d / %d / %d\n'
+            % tuple(datasize)
+        )
+    ui.writenoi18n(
+        b'full revision size (min/max/avg)     : %d / %d / %d\n'
+        % tuple(fullsize)
+    )
+    ui.writenoi18n(
+        b'inter-snapshot size (min/max/avg)    : %d / %d / %d\n'
+        % tuple(semisize)
+    )
+    for depth in sorted(snapsizedepth):
+        if depth == 0:
+            continue
+        ui.writenoi18n(
+            b'    level-%-3d (min/max/avg)          : %d / %d / %d\n'
+            % ((depth,) + tuple(snapsizedepth[depth]))
+        )
+    ui.writenoi18n(
+        b'delta size (min/max/avg)             : %d / %d / %d\n'
+        % tuple(deltasize)
+    )
+
+    if numdeltas > 0:
+        ui.write(b'\n')
+        fmt = pcfmtstr(numdeltas)
+        fmt2 = pcfmtstr(numdeltas, 4)
+        ui.writenoi18n(
+            b'deltas against prev  : ' + fmt % pcfmt(numprev, numdeltas)
+        )
+        if numprev > 0:
+            ui.writenoi18n(
+                b'    where prev = p1  : ' + fmt2 % pcfmt(nump1prev, numprev)
+            )
+            ui.writenoi18n(
+                b'    where prev = p2  : ' + fmt2 % pcfmt(nump2prev, numprev)
+            )
+            ui.writenoi18n(
+                b'    other-ancestor   : ' + fmt2 % pcfmt(numoprev, numprev)
+            )
+            ui.writenoi18n(
+                b'    unrelated        : ' + fmt2 % pcfmt(numoprev, numprev)
+            )
+        if gdelta:
+            ui.writenoi18n(
+                b'deltas against p1    : ' + fmt % pcfmt(nump1, numdeltas)
+            )
+            ui.writenoi18n(
+                b'deltas against p2    : ' + fmt % pcfmt(nump2, numdeltas)
+            )
+            ui.writenoi18n(
+                b'deltas against ancs  : '
+                + fmt % pcfmt(num_other_ancestors, numdeltas)
+            )
+            ui.writenoi18n(
+                b'deltas against other : '
+                + fmt % pcfmt(numother_nad, numdeltas)
+            )
+
+
+def debug_delta_find(ui, revlog, rev, base_rev=nodemod.nullrev):
+    """display the search process for a delta"""
+    deltacomputer = deltautil.deltacomputer(
+        revlog,
+        write_debug=ui.write,
+        debug_search=not ui.quiet,
+    )
+
+    node = revlog.node(rev)
+    p1r, p2r = revlog.parentrevs(rev)
+    p1 = revlog.node(p1r)
+    p2 = revlog.node(p2r)
+    full_text = revlog.revision(rev)
+    btext = [full_text]
+    textlen = len(btext[0])
+    cachedelta = None
+    flags = revlog.flags(rev)
+
+    if base_rev != nodemod.nullrev:
+        base_text = revlog.revision(base_rev)
+        delta = mdiff.textdiff(base_text, full_text)
+
+        cachedelta = (base_rev, delta, constants.DELTA_BASE_REUSE_TRY)
+        btext = [None]
+
+    revinfo = revlogutils.revisioninfo(
+        node,
+        p1,
+        p2,
+        btext,
+        textlen,
+        cachedelta,
+        flags,
+    )
+
+    fh = revlog._datafp()
+    deltacomputer.finddeltainfo(revinfo, fh, target_rev=rev)
+
+
+def _get_revlogs(repo, changelog: bool, manifest: bool, filelogs: bool):
+    """yield revlogs from this repository"""
+    if changelog:
+        yield repo.changelog
+
+    if manifest:
+        # XXX: Handle tree manifest
+        root_mf = repo.manifestlog.getstorage(b'')
+        assert not root_mf._treeondisk
+        yield root_mf._revlog
+
+    if filelogs:
+        files = set()
+        for rev in repo:
+            ctx = repo[rev]
+            files |= set(ctx.files())
+
+        for f in sorted(files):
+            yield repo.file(f)._revlog
+
+
+def debug_revlog_stats(
+    repo, fm, changelog: bool, manifest: bool, filelogs: bool
+):
+    """Format revlog statistics for debugging purposes
+
+    fm: the output formatter.
+    """
+    fm.plain(b'rev-count   data-size inl type      target \n')
+
+    for rlog in _get_revlogs(repo, changelog, manifest, filelogs):
+        fm.startitem()
+        nb_rev = len(rlog)
+        inline = rlog._inline
+        data_size = rlog._get_data_offset(nb_rev - 1)
+
+        target = rlog.target
+        revlog_type = b'unknown'
+        revlog_target = b''
+        if target[0] == constants.KIND_CHANGELOG:
+            revlog_type = b'changelog'
+        elif target[0] == constants.KIND_MANIFESTLOG:
+            revlog_type = b'manifest'
+            revlog_target = target[1]
+        elif target[0] == constants.KIND_FILELOG:
+            revlog_type = b'file'
+            revlog_target = target[1]
+
+        fm.write(b'revlog.rev-count', b'%9d', nb_rev)
+        fm.write(b'revlog.data-size', b'%12d', data_size)
+
+        fm.write(b'revlog.inline', b' %-3s', b'yes' if inline else b'no')
+        fm.write(b'revlog.type', b' %-9s', revlog_type)
+        fm.write(b'revlog.target', b' %s', revlog_target)
+
+        fm.plain(b'\n')

mercurial/revlogutils/deltas.py

@@ -20,6 +20,8 b' from .constants import ('
     COMP_MODE_DEFAULT,
     COMP_MODE_INLINE,
     COMP_MODE_PLAIN,
+    DELTA_BASE_REUSE_FORCE,
+    DELTA_BASE_REUSE_NO,
     KIND_CHANGELOG,
     KIND_FILELOG,
     KIND_MANIFESTLOG,
@@ -576,13 +578,20 b' def drop_u_compression(delta):'
     )
 
 
-def isgooddeltainfo(revlog, deltainfo, revinfo):
+def is_good_delta_info(revlog, deltainfo, revinfo):
     """Returns True if the given delta is good. Good means that it is within
     the disk span, disk size, and chain length bounds that we know to be
     performant."""
     if deltainfo is None:
         return False
 
+    if (
+        revinfo.cachedelta is not None
+        and deltainfo.base == revinfo.cachedelta[0]
+        and revinfo.cachedelta[2] == DELTA_BASE_REUSE_FORCE
+    ):
+        return True
+
     # - 'deltainfo.distance' is the distance from the base revision --
     #   bounding it limits the amount of I/O we need to do.
     # - 'deltainfo.compresseddeltalen' is the sum of the total size of
@@ -655,7 +664,16 b' def isgooddeltainfo(revlog, deltainfo, r'
 LIMIT_BASE2TEXT = 500
 
 
-def _candidategroups(revlog, textlen, p1, p2, cachedelta):
+def _candidategroups(
+    revlog,
+    textlen,
+    p1,
+    p2,
+    cachedelta,
+    excluded_bases=None,
+    target_rev=None,
+    snapshot_cache=None,
+):
     """Provides group of revision to be tested as delta base
 
     This top level function focus on emitting groups with unique and worthwhile
@@ -666,15 +684,31 b' def _candidategroups(revlog, textlen, p1'
         yield None
         return
 
+    if (
+        cachedelta is not None
+        and nullrev == cachedelta[0]
+        and cachedelta[2] == DELTA_BASE_REUSE_FORCE
+    ):
+        # instruction are to forcibly do a full snapshot
+        yield None
+        return
+
     deltalength = revlog.length
     deltaparent = revlog.deltaparent
     sparse = revlog._sparserevlog
     good = None
 
     deltas_limit = textlen * LIMIT_DELTA2TEXT
+    group_chunk_size = revlog._candidate_group_chunk_size
 
     tested = {nullrev}
-    candidates = _refinedgroups(revlog, p1, p2, cachedelta)
+    candidates = _refinedgroups(
+        revlog,
+        p1,
+        p2,
+        cachedelta,
+        snapshot_cache=snapshot_cache,
+    )
     while True:
         temptative = candidates.send(good)
         if temptative is None:
@@ -694,15 +728,37 b' def _candidategroups(revlog, textlen, p1'
             # filter out revision we tested already
             if rev in tested:
                 continue
-            tested.add(rev)
+
+            if (
+                cachedelta is not None
+                and rev == cachedelta[0]
+                and cachedelta[2] == DELTA_BASE_REUSE_FORCE
+            ):
+                # instructions are to forcibly consider/use this delta base
+                group.append(rev)
+                continue
+
+            # an higher authority deamed the base unworthy (e.g. censored)
+            if excluded_bases is not None and rev in excluded_bases:
+                tested.add(rev)
+                continue
+            # We are in some recomputation cases and that rev is too high in
+            # the revlog
+            if target_rev is not None and rev >= target_rev:
+                tested.add(rev)
+                continue
             # filter out delta base that will never produce good delta
             if deltas_limit < revlog.length(rev):
+                tested.add(rev)
                 continue
             if sparse and revlog.rawsize(rev) < (textlen // LIMIT_BASE2TEXT):
+                tested.add(rev)
                 continue
             # no delta for rawtext-changing revs (see "candelta" for why)
             if revlog.flags(rev) & REVIDX_RAWTEXT_CHANGING_FLAGS:
+                tested.add(rev)
                 continue
+
             # If we reach here, we are about to build and test a delta.
             # The delta building process will compute the chaininfo in all
             # case, since that computation is cached, it is fine to access it
@@ -710,9 +766,11 b' def _candidategroups(revlog, textlen, p1'
             chainlen, chainsize = revlog._chaininfo(rev)
             # if chain will be too long, skip base
             if revlog._maxchainlen and chainlen >= revlog._maxchainlen:
+                tested.add(rev)
                 continue
             # if chain already have too much data, skip base
             if deltas_limit < chainsize:
+                tested.add(rev)
                 continue
             if sparse and revlog.upperboundcomp is not None:
                 maxcomp = revlog.upperboundcomp
@@ -731,36 +789,46 b' def _candidategroups(revlog, textlen, p1'
                     snapshotlimit = textlen >> snapshotdepth
                     if snapshotlimit < lowestrealisticdeltalen:
                         # delta lower bound is larger than accepted upper bound
+                        tested.add(rev)
                         continue
 
                     # check the relative constraint on the delta size
                     revlength = revlog.length(rev)
                     if revlength < lowestrealisticdeltalen:
                         # delta probable lower bound is larger than target base
+                        tested.add(rev)
                         continue
 
             group.append(rev)
         if group:
-            # XXX: in the sparse revlog case, group can become large,
-            #      impacting performances. Some bounding or slicing mecanism
-            #      would help to reduce this impact.
-            good = yield tuple(group)
+            # When the size of the candidate group is big, it can result in a
+            # quite significant performance impact. To reduce this, we can send
+            # them in smaller batches until the new batch does not provide any
+            # improvements.
+            #
+            # This might reduce the overall efficiency of the compression in
+            # some corner cases, but that should also prevent very pathological
+            # cases from being an issue. (eg. 20 000 candidates).
+            #
+            # XXX note that the ordering of the group becomes important as it
+            # now impacts the final result. The current order is unprocessed
+            # and can be improved.
+            if group_chunk_size == 0:
+                tested.update(group)
+                good = yield tuple(group)
+            else:
+                prev_good = good
+                for start in range(0, len(group), group_chunk_size):
+                    sub_group = group[start : start + group_chunk_size]
+                    tested.update(sub_group)
+                    good = yield tuple(sub_group)
+                    if prev_good == good:
+                        break
+
     yield None
 
 
-def _findsnapshots(revlog, cache, start_rev):
-    """find snapshot from start_rev to tip"""
-    if util.safehasattr(revlog.index, b'findsnapshots'):
-        revlog.index.findsnapshots(cache, start_rev)
-    else:
-        deltaparent = revlog.deltaparent
-        issnapshot = revlog.issnapshot
-        for rev in revlog.revs(start_rev):
-            if issnapshot(rev):
-                cache[deltaparent(rev)].append(rev)
-
-
-def _refinedgroups(revlog, p1, p2, cachedelta):
+def _refinedgroups(revlog, p1, p2, cachedelta, snapshot_cache=None):
     good = None
     # First we try to reuse a the delta contained in the bundle.
     # (or from the source revlog)
@@ -768,15 +836,28 b' def _refinedgroups(revlog, p1, p2, cache'
     # This logic only applies to general delta repositories and can be disabled
     # through configuration. Disabling reuse source delta is useful when
     # we want to make sure we recomputed "optimal" deltas.
-    if cachedelta and revlog._generaldelta and revlog._lazydeltabase:
+    debug_info = None
+    if cachedelta is not None and cachedelta[2] > DELTA_BASE_REUSE_NO:
         # Assume what we received from the server is a good choice
         # build delta will reuse the cache
+        if debug_info is not None:
+            debug_info['cached-delta.tested'] += 1
         good = yield (cachedelta[0],)
         if good is not None:
+            if debug_info is not None:
+                debug_info['cached-delta.accepted'] += 1
             yield None
             return
-    snapshots = collections.defaultdict(list)
-    for candidates in _rawgroups(revlog, p1, p2, cachedelta, snapshots):
+    if snapshot_cache is None:
+        snapshot_cache = SnapshotCache()
+    groups = _rawgroups(
+        revlog,
+        p1,
+        p2,
+        cachedelta,
+        snapshot_cache,
+    )
+    for candidates in groups:
         good = yield candidates
         if good is not None:
             break
@@ -797,19 +878,22 b' def _refinedgroups(revlog, p1, p2, cache'
                 break
             good = yield (base,)
         # refine snapshot up
-        if not snapshots:
-            _findsnapshots(revlog, snapshots, good + 1)
+        if not snapshot_cache.snapshots:
+            snapshot_cache.update(revlog, good + 1)
         previous = None
         while good != previous:
             previous = good
-            children = tuple(sorted(c for c in snapshots[good]))
+            children = tuple(sorted(c for c in snapshot_cache.snapshots[good]))
             good = yield children
 
-    # we have found nothing
+    if debug_info is not None:
+        if good is None:
+            debug_info['no-solution'] += 1
+
     yield None
 
 
-def _rawgroups(revlog, p1, p2, cachedelta, snapshots=None):
+def _rawgroups(revlog, p1, p2, cachedelta, snapshot_cache=None):
     """Provides group of revision to be tested as delta base
 
     This lower level function focus on emitting delta theorically interresting
@@ -840,9 +924,9 b' def _rawgroups(revlog, p1, p2, cachedelt'
             yield parents
 
     if sparse and parents:
-        if snapshots is None:
-            # map: base-rev: snapshot-rev
-            snapshots = collections.defaultdict(list)
+        if snapshot_cache is None:
+            # map: base-rev: [snapshot-revs]
+            snapshot_cache = SnapshotCache()
         # See if we can use an existing snapshot in the parent chains to use as
         # a base for a new intermediate-snapshot
         #
@@ -856,7 +940,7 b' def _rawgroups(revlog, p1, p2, cachedelt'
                     break
                 parents_snaps[idx].add(s)
         snapfloor = min(parents_snaps[0]) + 1
-        _findsnapshots(revlog, snapshots, snapfloor)
+        snapshot_cache.update(revlog, snapfloor)
         # search for the highest "unrelated" revision
         #
         # Adding snapshots used by "unrelated" revision increase the odd we
@@ -879,14 +963,14 b' def _rawgroups(revlog, p1, p2, cachedelt'
             # chain.
             max_depth = max(parents_snaps.keys())
             chain = deltachain(other)
-            for idx, s in enumerate(chain):
+            for depth, s in enumerate(chain):
                 if s < snapfloor:
                     continue
-                if max_depth < idx:
+                if max_depth < depth:
                     break
                 if not revlog.issnapshot(s):
                     break
-                parents_snaps[idx].add(s)
+                parents_snaps[depth].add(s)
         # Test them as possible intermediate snapshot base
         # We test them from highest to lowest level. High level one are more
         # likely to result in small delta
@@ -894,7 +978,7 b' def _rawgroups(revlog, p1, p2, cachedelt'
         for idx, snaps in sorted(parents_snaps.items(), reverse=True):
             siblings = set()
             for s in snaps:
-                siblings.update(snapshots[s])
+                siblings.update(snapshot_cache.snapshots[s])
             # Before considering making a new intermediate snapshot, we check
             # if an existing snapshot, children of base we consider, would be
             # suitable.
@@ -922,7 +1006,8 b' def _rawgroups(revlog, p1, p2, cachedelt'
         # revisions instead of starting our own. Without such re-use,
         # topological branches would keep reopening new full chains. Creating
         # more and more snapshot as the repository grow.
-        yield tuple(snapshots[nullrev])
+        full = [r for r in snapshot_cache.snapshots[nullrev] if snapfloor <= r]
+        yield tuple(sorted(full))
 
     if not sparse:
         # other approach failed try against prev to hopefully save us a
@@ -930,11 +1015,74 b' def _rawgroups(revlog, p1, p2, cachedelt'
         yield (prev,)
 
 
+class SnapshotCache:
+    __slots__ = ('snapshots', '_start_rev', '_end_rev')
+
+    def __init__(self):
+        self.snapshots = collections.defaultdict(set)
+        self._start_rev = None
+        self._end_rev = None
+
+    def update(self, revlog, start_rev=0):
+        """find snapshots from start_rev to tip"""
+        nb_revs = len(revlog)
+        end_rev = nb_revs - 1
+        if start_rev > end_rev:
+            return  # range is empty
+
+        if self._start_rev is None:
+            assert self._end_rev is None
+            self._update(revlog, start_rev, end_rev)
+        elif not (self._start_rev <= start_rev and end_rev <= self._end_rev):
+            if start_rev < self._start_rev:
+                self._update(revlog, start_rev, self._start_rev - 1)
+            if self._end_rev < end_rev:
+                self._update(revlog, self._end_rev + 1, end_rev)
+
+        if self._start_rev is None:
+            assert self._end_rev is None
+            self._end_rev = end_rev
+            self._start_rev = start_rev
+        else:
+            self._start_rev = min(self._start_rev, start_rev)
+            self._end_rev = max(self._end_rev, end_rev)
+        assert self._start_rev <= self._end_rev, (
+            self._start_rev,
+            self._end_rev,
+        )
+
+    def _update(self, revlog, start_rev, end_rev):
+        """internal method that actually do update content"""
+        assert self._start_rev is None or (
+            start_rev < self._start_rev or start_rev > self._end_rev
+        ), (self._start_rev, self._end_rev, start_rev, end_rev)
+        assert self._start_rev is None or (
+            end_rev < self._start_rev or end_rev > self._end_rev
+        ), (self._start_rev, self._end_rev, start_rev, end_rev)
+        cache = self.snapshots
+        if util.safehasattr(revlog.index, b'findsnapshots'):
+            revlog.index.findsnapshots(cache, start_rev, end_rev)
+        else:
+            deltaparent = revlog.deltaparent
+            issnapshot = revlog.issnapshot
+            for rev in revlog.revs(start_rev, end_rev):
+                if issnapshot(rev):
+                    cache[deltaparent(rev)].add(rev)
+
+
 class deltacomputer:
-    def __init__(self, revlog, write_debug=None, debug_search=False):
+    def __init__(
+        self,
+        revlog,
+        write_debug=None,
+        debug_search=False,
+        debug_info=None,
+    ):
         self.revlog = revlog
         self._write_debug = write_debug
         self._debug_search = debug_search
+        self._debug_info = debug_info
+        self._snapshot_cache = SnapshotCache()
 
     def buildtext(self, revinfo, fh):
         """Builds a fulltext version of a revision
@@ -998,7 +1146,7 b' class deltacomputer:'
                 snapshotdepth = len(revlog._deltachain(deltabase)[0])
         delta = None
         if revinfo.cachedelta:
-            cachebase, cachediff = revinfo.cachedelta
+            cachebase = revinfo.cachedelta[0]
             # check if the diff still apply
             currentbase = cachebase
             while (
@@ -1103,11 +1251,14 b' class deltacomputer:'
         if revinfo.flags & REVIDX_RAWTEXT_CHANGING_FLAGS:
             return self._fullsnapshotinfo(fh, revinfo, target_rev)
 
-        if self._write_debug is not None:
+        gather_debug = (
+            self._write_debug is not None or self._debug_info is not None
+        )
+        debug_search = self._write_debug is not None and self._debug_search
+
+        if gather_debug:
             start = util.timer()
 
-        debug_search = self._write_debug is not None and self._debug_search
-
         # count the number of different delta we tried (for debug purpose)
         dbg_try_count = 0
         # count the number of "search round" we did. (for debug purpose)
@@ -1122,7 +1273,7 b' class deltacomputer:'
         deltainfo = None
         p1r, p2r = revlog.rev(p1), revlog.rev(p2)
 
-        if self._write_debug is not None:
+        if gather_debug:
             if p1r != nullrev:
                 p1_chain_len = revlog._chaininfo(p1r)[0]
             else:
@@ -1137,7 +1288,14 b' class deltacomputer:'
             self._write_debug(msg)
 
         groups = _candidategroups(
-            self.revlog, revinfo.textlen, p1r, p2r, cachedelta
+            self.revlog,
+            revinfo.textlen,
+            p1r,
+            p2r,
+            cachedelta,
+            excluded_bases,
+            target_rev,
+            snapshot_cache=self._snapshot_cache,
         )
         candidaterevs = next(groups)
         while candidaterevs is not None:
@@ -1147,7 +1305,13 b' class deltacomputer:'
                 if deltainfo is not None:
                     prev = deltainfo.base
 
-                if p1 in candidaterevs or p2 in candidaterevs:
+                if (
+                    cachedelta is not None
+                    and len(candidaterevs) == 1
+                    and cachedelta[0] in candidaterevs
+                ):
+                    round_type = b"cached-delta"
+                elif p1 in candidaterevs or p2 in candidaterevs:
                     round_type = b"parents"
                 elif prev is not None and all(c < prev for c in candidaterevs):
                     round_type = b"refine-down"
@@ -1195,16 +1359,7 b' class deltacomputer:'
                     msg = b"DBG-DELTAS-SEARCH:     base=%d\n"
                     msg %= self.revlog.deltaparent(candidaterev)
                     self._write_debug(msg)
-                if candidaterev in excluded_bases:
-                    if debug_search:
-                        msg = b"DBG-DELTAS-SEARCH:     EXCLUDED\n"
-                        self._write_debug(msg)
-                    continue
-                if candidaterev >= target_rev:
-                    if debug_search:
-                        msg = b"DBG-DELTAS-SEARCH:     TOO-HIGH\n"
-                        self._write_debug(msg)
-                    continue
+
                 dbg_try_count += 1
 
                 if debug_search:
@@ -1216,7 +1371,7 b' class deltacomputer:'
                     msg %= delta_end - delta_start
                     self._write_debug(msg)
                 if candidatedelta is not None:
-                    if isgooddeltainfo(self.revlog, candidatedelta, revinfo):
+                    if is_good_delta_info(self.revlog, candidatedelta, revinfo):
                         if debug_search:
                             msg = b"DBG-DELTAS-SEARCH:     DELTA: length=%d (GOOD)\n"
                             msg %= candidatedelta.deltalen
@@ -1244,12 +1399,28 b' class deltacomputer:'
         else:
             dbg_type = b"delta"
 
-        if self._write_debug is not None:
+        if gather_debug:
             end = util.timer()
+            if dbg_type == b'full':
+                used_cached = (
+                    cachedelta is not None
+                    and dbg_try_rounds == 0
+                    and dbg_try_count == 0
+                    and cachedelta[0] == nullrev
+                )
+            else:
+                used_cached = (
+                    cachedelta is not None
+                    and dbg_try_rounds == 1
+                    and dbg_try_count == 1
+                    and deltainfo.base == cachedelta[0]
+                )
             dbg = {
                 'duration': end - start,
                 'revision': target_rev,
+                'delta-base': deltainfo.base,  # pytype: disable=attribute-error
                 'search_round_count': dbg_try_rounds,
+                'using-cached-base': used_cached,
                 'delta_try_count': dbg_try_count,
                 'type': dbg_type,
                 'p1-chain-len': p1_chain_len,
@@ -1279,31 +1450,39 b' class deltacomputer:'
                     target_revlog += b'%s:' % target_key
             dbg['target-revlog'] = target_revlog
 
-            msg = (
-                b"DBG-DELTAS:"
-                b" %-12s"
-                b" rev=%d:"
-                b" search-rounds=%d"
-                b" try-count=%d"
-                b" - delta-type=%-6s"
-                b" snap-depth=%d"
-                b" - p1-chain-length=%d"
-                b" p2-chain-length=%d"
-                b" - duration=%f"
-                b"\n"
-            )
-            msg %= (
-                dbg["target-revlog"],
-                dbg["revision"],
-                dbg["search_round_count"],
-                dbg["delta_try_count"],
-                dbg["type"],
-                dbg["snapshot-depth"],
-                dbg["p1-chain-len"],
-                dbg["p2-chain-len"],
-                dbg["duration"],
-            )
-            self._write_debug(msg)
+            if self._debug_info is not None:
+                self._debug_info.append(dbg)
+
+            if self._write_debug is not None:
+                msg = (
+                    b"DBG-DELTAS:"
+                    b" %-12s"
+                    b" rev=%d:"
+                    b" delta-base=%d"
+                    b" is-cached=%d"
+                    b" - search-rounds=%d"
+                    b" try-count=%d"
+                    b" - delta-type=%-6s"
+                    b" snap-depth=%d"
+                    b" - p1-chain-length=%d"
+                    b" p2-chain-length=%d"
+                    b" - duration=%f"
+                    b"\n"
+                )
+                msg %= (
+                    dbg["target-revlog"],
+                    dbg["revision"],
+                    dbg["delta-base"],
+                    dbg["using-cached-base"],
+                    dbg["search_round_count"],
+                    dbg["delta_try_count"],
+                    dbg["type"],
+                    dbg["snapshot-depth"],
+                    dbg["p1-chain-len"],
+                    dbg["p2-chain-len"],
+                    dbg["duration"],
+                )
+                self._write_debug(msg)
         return deltainfo
 
 

mercurial/revlogutils/docket.py

@@ -90,7 +90,7 b' if stable_docket_file:'
 # * 8 bytes: pending size of data
 # * 8 bytes: pending size of sidedata
 # * 1 bytes: default compression header
-S_HEADER = struct.Struct(constants.INDEX_HEADER_FMT + b'BBBBBBLLLLLLc')
+S_HEADER = struct.Struct(constants.INDEX_HEADER_FMT + b'BBBBBBQQQQQQc')
 # * 1 bytes: size of index uuid
 # * 8 bytes: size of file
 S_OLD_UID = struct.Struct('>BL')

mercurial/revset.py

@@ -1868,13 +1868,12 b' def outgoing(repo, subset, x):'
         dests = []
     missing = set()
     for path in urlutil.get_push_paths(repo, repo.ui, dests):
-        dest = path.pushloc or path.loc
         branches = path.branch, []
 
         revs, checkout = hg.addbranchrevs(repo, repo, branches, [])
         if revs:
             revs = [repo.lookup(rev) for rev in revs]
-        other = hg.peer(repo, {}, dest)
+        other = hg.peer(repo, {}, path)
         try:
             with repo.ui.silent():
                 outgoing = discovery.findcommonoutgoing(
@@ -2130,11 +2129,9 b' def remote(repo, subset, x):'
         dest = getstring(l[1], _(b"remote requires a repository path"))
     if not dest:
         dest = b'default'
-    dest, branches = urlutil.get_unique_pull_path(
-        b'remote', repo, repo.ui, dest
-    )
-
-    other = hg.peer(repo, {}, dest)
+    path = urlutil.get_unique_pull_path_obj(b'remote', repo.ui, dest)
+
+    other = hg.peer(repo, {}, path)
     n = other.lookup(q)
     if n in repo:
         r = repo[n].rev()

mercurial/scmposix.py

@@ -4,6 +4,11 b' import fcntl'
 import os
 import sys
 
+from typing import (
+    List,
+    Tuple,
+)
+
 from .pycompat import getattr
 from . import (
     encoding,
@@ -11,6 +16,9 b' from . import ('
     util,
 )
 
+if pycompat.TYPE_CHECKING:
+    from . import ui as uimod
+
 # BSD 'more' escapes ANSI color sequences by default. This can be disabled by
 # $MORE variable, but there's no compatible option with Linux 'more'. Given
 # OS X is widely used and most modern Unix systems would have 'less', setting
@@ -18,7 +26,7 b' from . import ('
 fallbackpager = b'less'
 
 
-def _rcfiles(path):
+def _rcfiles(path: bytes) -> List[bytes]:
     rcs = [os.path.join(path, b'hgrc')]
     rcdir = os.path.join(path, b'hgrc.d')
     try:
@@ -34,7 +42,7 b' def _rcfiles(path):'
     return rcs
 
 
-def systemrcpath():
+def systemrcpath() -> List[bytes]:
     path = []
     if pycompat.sysplatform == b'plan9':
         root = b'lib/mercurial'
@@ -49,7 +57,7 b' def systemrcpath():'
     return path
 
 
-def userrcpath():
+def userrcpath() -> List[bytes]:
     if pycompat.sysplatform == b'plan9':
         return [encoding.environ[b'home'] + b'/lib/hgrc']
     elif pycompat.isdarwin:
@@ -65,7 +73,7 b' def userrcpath():'
         ]
 
 
-def termsize(ui):
+def termsize(ui: "uimod.ui") -> Tuple[int, int]:
     try:
         import termios
 
@@ -88,7 +96,7 b' def termsize(ui):'
         except ValueError:
             pass
         except IOError as e:
-            if e[0] == errno.EINVAL:  # pytype: disable=unsupported-operands
+            if e.errno == errno.EINVAL:
                 pass
             else:
                 raise

mercurial/scmutil.py

@@ -1219,7 +1219,7 b' def cleanupnodes('
                 )
 
 
-def addremove(repo, matcher, prefix, uipathfn, opts=None):
+def addremove(repo, matcher, prefix, uipathfn, opts=None, open_tr=None):
     if opts is None:
         opts = {}
     m = matcher
@@ -1279,7 +1279,9 b' def addremove(repo, matcher, prefix, uip'
         repo, m, added + unknown, removed + deleted, similarity, uipathfn
     )
 
-    if not dry_run:
+    if not dry_run and (unknown or forgotten or deleted or renames):
+        if open_tr is not None:
+            open_tr()
         _markchanges(repo, unknown + forgotten, deleted, renames)
 
     for f in rejected:
@@ -1863,7 +1865,12 b' def gdinitconfig(ui):'
 
 
 def gddeltaconfig(ui):
-    """helper function to know if incoming delta should be optimised"""
+    """helper function to know if incoming deltas should be optimized
+
+    The `format.generaldelta` config is an old form of the config that also
+    implies that incoming delta-bases should be never be trusted. This function
+    exists for this purpose.
+    """
     # experimental config: format.generaldelta
     return ui.configbool(b'format', b'generaldelta')
 

mercurial/scmwindows.py

@@ -1,4 +1,10 b''
 import os
+import winreg  # pytype: disable=import-error
+
+from typing import (
+    List,
+    Tuple,
+)
 
 from . import (
     encoding,
@@ -7,19 +13,14 b' from . import ('
     win32,
 )
 
-try:
-    import _winreg as winreg  # pytype: disable=import-error
-
-    winreg.CloseKey
-except ImportError:
-    # py2 only
-    import winreg  # pytype: disable=import-error
+if pycompat.TYPE_CHECKING:
+    from . import ui as uimod
 
 # MS-DOS 'more' is the only pager available by default on Windows.
 fallbackpager = b'more'
 
 
-def systemrcpath():
+def systemrcpath() -> List[bytes]:
     '''return default os-specific hgrc search path'''
     rcpath = []
     filename = win32.executablepath()
@@ -27,7 +28,7 b' def systemrcpath():'
     progrc = os.path.join(os.path.dirname(filename), b'mercurial.ini')
     rcpath.append(progrc)
 
-    def _processdir(progrcd):
+    def _processdir(progrcd: bytes) -> None:
         if os.path.isdir(progrcd):
             for f, kind in sorted(util.listdir(progrcd)):
                 if f.endswith(b'.rc'):
@@ -68,7 +69,7 b' def systemrcpath():'
     return rcpath
 
 
-def userrcpath():
+def userrcpath() -> List[bytes]:
     '''return os-specific hgrc search path to the user dir'''
     home = _legacy_expanduser(b'~')
     path = [os.path.join(home, b'mercurial.ini'), os.path.join(home, b'.hgrc')]
@@ -79,7 +80,7 b' def userrcpath():'
     return path
 
 
-def _legacy_expanduser(path):
+def _legacy_expanduser(path: bytes) -> bytes:
     """Expand ~ and ~user constructs in the pre 3.8 style"""
 
     # Python 3.8+ changed the expansion of '~' from HOME to USERPROFILE.  See
@@ -111,5 +112,5 b' def _legacy_expanduser(path):'
     return userhome + path[i:]
 
 
-def termsize(ui):
+def termsize(ui: "uimod.ui") -> Tuple[int, int]:
     return win32.termsize()

mercurial/shelve.py

@@ -247,6 +247,14 b' class Shelf:'
         for ext in shelvefileextensions:
             self.vfs.tryunlink(self.name + b'.' + ext)
 
+    def changed_files(self, ui, repo):
+        try:
+            ctx = repo.unfiltered()[self.readinfo()[b'node']]
+            return ctx.files()
+        except (FileNotFoundError, error.RepoLookupError):
+            filename = self.vfs.join(self.name + b'.patch')
+            return patch.changedfiles(ui, repo, filename)
+
 
 def _optimized_match(repo, node):
     """
@@ -424,10 +432,26 b' def _restoreactivebookmark(repo, mark):'
 
 def _aborttransaction(repo, tr):
     """Abort current transaction for shelve/unshelve, but keep dirstate"""
-    dirstatebackupname = b'dirstate.shelve'
-    repo.dirstate.savebackup(None, dirstatebackupname)
-    tr.abort()
-    repo.dirstate.restorebackup(None, dirstatebackupname)
+    # disable the transaction invalidation of the dirstate, to preserve the
+    # current change in memory.
+    ds = repo.dirstate
+    # The assert below check that nobody else did such wrapping.
+    #
+    # These is not such other wrapping currently, but if someone try to
+    # implement one in the future, this will explicitly break here instead of
+    # misbehaving in subtle ways.
+    assert 'invalidate' not in vars(ds)
+    try:
+        # note : we could simply disable the transaction abort callback, but
+        # other code also tries to rollback and invalidate this.
+        ds.invalidate = lambda: None
+        tr.abort()
+    finally:
+        del ds.invalidate
+    # manually write the change in memory since we can no longer rely on the
+    # transaction to do so.
+    assert repo.currenttransaction() is None
+    repo.dirstate.write(None)
 
 
 def getshelvename(repo, parent, opts):
@@ -599,7 +623,8 b' def _docreatecmd(ui, repo, pats, opts):'
         activebookmark = _backupactivebookmark(repo)
         extra = {b'internal': b'shelve'}
         if includeunknown:
-            _includeunknownfiles(repo, pats, opts, extra)
+            with repo.dirstate.changing_files(repo):
+                _includeunknownfiles(repo, pats, opts, extra)
 
         if _iswctxonnewbranch(repo) and not _isbareshelve(pats, opts):
             # In non-bare shelve we don't store newly created branch
@@ -629,7 +654,7 b' def _docreatecmd(ui, repo, pats, opts):'
 
         ui.status(_(b'shelved as %s\n') % name)
         if opts[b'keep']:
-            with repo.dirstate.parentchange():
+            with repo.dirstate.changing_parents(repo):
                 scmutil.movedirstate(repo, parent, match)
         else:
             hg.update(repo, parent.node())
@@ -854,18 +879,18 b' def unshelvecontinue(ui, repo, state, op'
         shelvectx = repo[state.parents[1]]
         pendingctx = state.pendingctx
 
-        with repo.dirstate.parentchange():
+        with repo.dirstate.changing_parents(repo):
             repo.setparents(state.pendingctx.node(), repo.nullid)
             repo.dirstate.write(repo.currenttransaction())
 
         targetphase = _target_phase(repo)
         overrides = {(b'phases', b'new-commit'): targetphase}
         with repo.ui.configoverride(overrides, b'unshelve'):
-            with repo.dirstate.parentchange():
+            with repo.dirstate.changing_parents(repo):
                 repo.setparents(state.parents[0], repo.nullid)
-                newnode, ispartialunshelve = _createunshelvectx(
-                    ui, repo, shelvectx, basename, interactive, opts
-                )
+            newnode, ispartialunshelve = _createunshelvectx(
+                ui, repo, shelvectx, basename, interactive, opts
+            )
 
         if newnode is None:
             shelvectx = state.pendingctx
@@ -1060,11 +1085,11 b' def _rebaserestoredcommit('
             )
             raise error.ConflictResolutionRequired(b'unshelve')
 
-        with repo.dirstate.parentchange():
+        with repo.dirstate.changing_parents(repo):
             repo.setparents(tmpwctx.node(), repo.nullid)
-            newnode, ispartialunshelve = _createunshelvectx(
-                ui, repo, shelvectx, basename, interactive, opts
-            )
+        newnode, ispartialunshelve = _createunshelvectx(
+            ui, repo, shelvectx, basename, interactive, opts
+        )
 
         if newnode is None:
             shelvectx = tmpwctx
@@ -1210,7 +1235,8 b' def _dounshelve(ui, repo, basename, opts'
         restorebranch(ui, repo, branchtorestore)
         shelvedstate.clear(repo)
         _finishunshelve(repo, oldtiprev, tr, activebookmark)
-        _forgetunknownfiles(repo, shelvectx, addedbefore)
+        with repo.dirstate.changing_files(repo):
+            _forgetunknownfiles(repo, shelvectx, addedbefore)
         if not ispartialunshelve:
             unshelvecleanup(ui, repo, basename, opts)
     finally:

mercurial/simplemerge.py

@@ -512,6 +512,8 b' def simplemerge('
     conflicts = False
     if mode == b'union':
         lines = _resolve(m3, (1, 2))
+    elif mode == b'union-other-first':
+        lines = _resolve(m3, (2, 1))
     elif mode == b'local':
         lines = _resolve(m3, (1,))
     elif mode == b'other':

mercurial/sparse.py

@@ -451,7 +451,7 b' def filterupdatesactions(repo, wctx, mct'
                     message,
                 )
 
-        with repo.dirstate.parentchange():
+        with repo.dirstate.changing_parents(repo):
             mergemod.applyupdates(
                 repo,
                 tmresult,
@@ -655,7 +655,7 b' def clearrules(repo, force=False):'
     The remaining sparse config only has profiles, if defined. The working
     directory is refreshed, as needed.
     """
-    with repo.wlock(), repo.dirstate.parentchange():
+    with repo.wlock(), repo.dirstate.changing_parents(repo):
         raw = repo.vfs.tryread(b'sparse')
         includes, excludes, profiles = parseconfig(repo.ui, raw, b'sparse')
 
@@ -671,7 +671,7 b' def importfromfiles(repo, opts, paths, f'
     The updated sparse config is written out and the working directory
     is refreshed, as needed.
     """
-    with repo.wlock(), repo.dirstate.parentchange():
+    with repo.wlock(), repo.dirstate.changing_parents(repo):
         # read current configuration
         raw = repo.vfs.tryread(b'sparse')
         includes, excludes, profiles = parseconfig(repo.ui, raw, b'sparse')
@@ -730,7 +730,7 b' def updateconfig('
 
     The new config is written out and a working directory refresh is performed.
     """
-    with repo.wlock(), repo.lock(), repo.dirstate.parentchange():
+    with repo.wlock(), repo.lock(), repo.dirstate.changing_parents(repo):
         raw = repo.vfs.tryread(b'sparse')
         oldinclude, oldexclude, oldprofiles = parseconfig(
             repo.ui, raw, b'sparse'

mercurial/sshpeer.py

@@ -372,7 +372,7 b' def _performhandshake(ui, stdin, stdout,'
 
 class sshv1peer(wireprotov1peer.wirepeer):
     def __init__(
-        self, ui, url, proc, stdin, stdout, stderr, caps, autoreadstderr=True
+        self, ui, path, proc, stdin, stdout, stderr, caps, autoreadstderr=True
     ):
         """Create a peer from an existing SSH connection.
 
@@ -383,8 +383,7 b' class sshv1peer(wireprotov1peer.wirepeer'
         ``autoreadstderr`` denotes whether to automatically read from
         stderr and to forward its output.
         """
-        self._url = url
-        self.ui = ui
+        super().__init__(ui, path=path)
         # self._subprocess is unused. Keeping a handle on the process
         # holds a reference and prevents it from being garbage collected.
         self._subprocess = proc
@@ -411,14 +410,11 b' class sshv1peer(wireprotov1peer.wirepeer'
     # Begin of ipeerconnection interface.
 
     def url(self):
-        return self._url
+        return self.path.loc
 
     def local(self):
         return None
 
-    def peer(self):
-        return self
-
     def canpush(self):
         return True
 
@@ -610,16 +606,16 b' def makepeer(ui, path, proc, stdin, stdo'
         )
 
 
-def instance(ui, path, create, intents=None, createopts=None):
+def make_peer(ui, path, create, intents=None, createopts=None):
     """Create an SSH peer.
 
     The returned object conforms to the ``wireprotov1peer.wirepeer`` interface.
     """
-    u = urlutil.url(path, parsequery=False, parsefragment=False)
+    u = urlutil.url(path.loc, parsequery=False, parsefragment=False)
     if u.scheme != b'ssh' or not u.host or u.path is None:
         raise error.RepoError(_(b"couldn't parse location %s") % path)
 
-    urlutil.checksafessh(path)
+    urlutil.checksafessh(path.loc)
 
     if u.passwd is not None:
         raise error.RepoError(_(b'password in URL not supported'))

mercurial/statichttprepo.py

@@ -225,6 +225,7 b' class statichttprepository('
         self.encodepats = None
         self.decodepats = None
         self._transref = None
+        self._dirstate = None
 
     def _restrictcapabilities(self, caps):
         caps = super(statichttprepository, self)._restrictcapabilities(caps)
@@ -236,8 +237,8 b' class statichttprepository('
     def local(self):
         return False
 
-    def peer(self):
-        return statichttppeer(self)
+    def peer(self, path=None):
+        return statichttppeer(self, path=path)
 
     def wlock(self, wait=True):
         raise error.LockUnavailable(
@@ -259,7 +260,8 b' class statichttprepository('
         pass  # statichttprepository are read only
 
 
-def instance(ui, path, create, intents=None, createopts=None):
+def make_peer(ui, path, create, intents=None, createopts=None):
     if create:
         raise error.Abort(_(b'cannot create new static-http repository'))
-    return statichttprepository(ui, path[7:])
+    url = path.loc[7:]
+    return statichttprepository(ui, url).peer(path=path)

mercurial/statprof.py

@@ -1049,7 +1049,7 b' def main(argv=None):'
     # process options
     try:
         opts, args = pycompat.getoptb(
-            sys.argv[optstart:],
+            pycompat.sysargv[optstart:],
             b"hl:f:o:p:",
             [b"help", b"limit=", b"file=", b"output-file=", b"script-path="],
         )

mercurial/strip.py

@@ -241,31 +241,32 b' def debugstrip(ui, repo, *revs, **opts):'
 
         revs = sorted(rootnodes)
         if update and opts.get(b'keep'):
-            urev = _findupdatetarget(repo, revs)
-            uctx = repo[urev]
+            with repo.dirstate.changing_parents(repo):
+                urev = _findupdatetarget(repo, revs)
+                uctx = repo[urev]
 
-            # only reset the dirstate for files that would actually change
-            # between the working context and uctx
-            descendantrevs = repo.revs(b"only(., %d)", uctx.rev())
-            changedfiles = []
-            for rev in descendantrevs:
-                # blindly reset the files, regardless of what actually changed
-                changedfiles.extend(repo[rev].files())
+                # only reset the dirstate for files that would actually change
+                # between the working context and uctx
+                descendantrevs = repo.revs(b"only(., %d)", uctx.rev())
+                changedfiles = []
+                for rev in descendantrevs:
+                    # blindly reset the files, regardless of what actually changed
+                    changedfiles.extend(repo[rev].files())
 
-            # reset files that only changed in the dirstate too
-            dirstate = repo.dirstate
-            dirchanges = [
-                f for f in dirstate if not dirstate.get_entry(f).maybe_clean
-            ]
-            changedfiles.extend(dirchanges)
+                # reset files that only changed in the dirstate too
+                dirstate = repo.dirstate
+                dirchanges = [
+                    f for f in dirstate if not dirstate.get_entry(f).maybe_clean
+                ]
+                changedfiles.extend(dirchanges)
 
-            repo.dirstate.rebuild(urev, uctx.manifest(), changedfiles)
-            repo.dirstate.write(repo.currenttransaction())
+                repo.dirstate.rebuild(urev, uctx.manifest(), changedfiles)
+                repo.dirstate.write(repo.currenttransaction())
 
-            # clear resolve state
-            mergestatemod.mergestate.clean(repo)
+                # clear resolve state
+                mergestatemod.mergestate.clean(repo)
 
-            update = False
+                update = False
 
         strip(
             ui,

mercurial/subrepo.py

@@ -569,9 +569,20 b' class hgsubrepo(abstractsubrepo):'
 
     @annotatesubrepoerror
     def add(self, ui, match, prefix, uipathfn, explicitonly, **opts):
-        return cmdutil.add(
-            ui, self._repo, match, prefix, uipathfn, explicitonly, **opts
-        )
+        # XXX Ideally, we could let the caller take the `changing_files`
+        # context.  However this is not an abstraction that make sense for
+        # other repository types, and leaking that details purely related to
+        # dirstate seems unfortunate. So for now the context will be used here.
+        with self._repo.wlock(), self._repo.dirstate.changing_files(self._repo):
+            return cmdutil.add(
+                ui,
+                self._repo,
+                match,
+                prefix,
+                uipathfn,
+                explicitonly,
+                **opts,
+            )
 
     @annotatesubrepoerror
     def addremove(self, m, prefix, uipathfn, opts):
@@ -580,7 +591,18 b' class hgsubrepo(abstractsubrepo):'
         # be used to process sibling subrepos however.
         opts = copy.copy(opts)
         opts[b'subrepos'] = True
-        return scmutil.addremove(self._repo, m, prefix, uipathfn, opts)
+        # XXX Ideally, we could let the caller take the `changing_files`
+        # context.  However this is not an abstraction that make sense for
+        # other repository types, and leaking that details purely related to
+        # dirstate seems unfortunate. So for now the context will be used here.
+        with self._repo.wlock(), self._repo.dirstate.changing_files(self._repo):
+            return scmutil.addremove(
+                self._repo,
+                m,
+                prefix,
+                uipathfn,
+                opts,
+            )
 
     @annotatesubrepoerror
     def cat(self, match, fm, fntemplate, prefix, **opts):
@@ -621,7 +643,7 b' class hgsubrepo(abstractsubrepo):'
                 match,
                 prefix=prefix,
                 listsubrepos=True,
-                **opts
+                **opts,
             )
         except error.RepoLookupError as inst:
             self.ui.warn(
@@ -946,16 +968,21 b' class hgsubrepo(abstractsubrepo):'
 
     @annotatesubrepoerror
     def forget(self, match, prefix, uipathfn, dryrun, interactive):
-        return cmdutil.forget(
-            self.ui,
-            self._repo,
-            match,
-            prefix,
-            uipathfn,
-            True,
-            dryrun=dryrun,
-            interactive=interactive,
-        )
+        # XXX Ideally, we could let the caller take the `changing_files`
+        # context.  However this is not an abstraction that make sense for
+        # other repository types, and leaking that details purely related to
+        # dirstate seems unfortunate. So for now the context will be used here.
+        with self._repo.wlock(), self._repo.dirstate.changing_files(self._repo):
+            return cmdutil.forget(
+                self.ui,
+                self._repo,
+                match,
+                prefix,
+                uipathfn,
+                True,
+                dryrun=dryrun,
+                interactive=interactive,
+            )
 
     @annotatesubrepoerror
     def removefiles(
@@ -969,17 +996,22 b' class hgsubrepo(abstractsubrepo):'
         dryrun,
         warnings,
     ):
-        return cmdutil.remove(
-            self.ui,
-            self._repo,
-            matcher,
-            prefix,
-            uipathfn,
-            after,
-            force,
-            subrepos,
-            dryrun,
-        )
+        # XXX Ideally, we could let the caller take the `changing_files`
+        # context.  However this is not an abstraction that make sense for
+        # other repository types, and leaking that details purely related to
+        # dirstate seems unfortunate. So for now the context will be used here.
+        with self._repo.wlock(), self._repo.dirstate.changing_files(self._repo):
+            return cmdutil.remove(
+                self.ui,
+                self._repo,
+                matcher,
+                prefix,
+                uipathfn,
+                after,
+                force,
+                subrepos,
+                dryrun,
+            )
 
     @annotatesubrepoerror
     def revert(self, substate, *pats, **opts):
@@ -1009,7 +1041,12 b' class hgsubrepo(abstractsubrepo):'
             pats = [b'set:modified()']
         else:
             pats = []
-        cmdutil.revert(self.ui, self._repo, ctx, *pats, **opts)
+        # XXX Ideally, we could let the caller take the `changing_files`
+        # context.  However this is not an abstraction that make sense for
+        # other repository types, and leaking that details purely related to
+        # dirstate seems unfortunate. So for now the context will be used here.
+        with self._repo.wlock(), self._repo.dirstate.changing_files(self._repo):
+            cmdutil.revert(self.ui, self._repo, ctx, *pats, **opts)
 
     def shortid(self, revid):
         return revid[:12]
@@ -1123,7 +1160,7 b' class svnsubrepo(abstractsubrepo):'
             stdout=subprocess.PIPE,
             stderr=subprocess.PIPE,
             env=procutil.tonativeenv(env),
-            **extrakw
+            **extrakw,
         )
         stdout, stderr = map(util.fromnativeeol, p.communicate())
         stderr = stderr.strip()
@@ -1488,7 +1525,7 b' class gitsubrepo(abstractsubrepo):'
             close_fds=procutil.closefds,
             stdout=subprocess.PIPE,
             stderr=errpipe,
-            **extrakw
+            **extrakw,
         )
         if stream:
             return p.stdout, None

mercurial/tags.py

@@ -664,8 +664,9 b' def _tag('
 
     repo.invalidatecaches()
 
-    if b'.hgtags' not in repo.dirstate:
-        repo[None].add([b'.hgtags'])
+    with repo.dirstate.changing_files(repo):
+        if b'.hgtags' not in repo.dirstate:
+            repo[None].add([b'.hgtags'])
 
     m = matchmod.exact([b'.hgtags'])
     tagnode = repo.commit(

mercurial/templater.py

@@ -177,10 +177,17 b' def tokenize(program, start, end, term=N'
             quote = program[pos : pos + 2]
             s = pos = pos + 2
             while pos < end:  # find closing escaped quote
+                # pycompat.bytestr (and bytes) both have .startswith() that
+                # takes an optional start and an optional end, but pytype thinks
+                # it only takes 2 args.
+
+                # pytype: disable=wrong-arg-count
                 if program.startswith(b'\\\\\\', pos, end):
                     pos += 4  # skip over double escaped characters
                     continue
                 if program.startswith(quote, pos, end):
+                    # pytype: enable=wrong-arg-count
+
                     # interpret as if it were a part of an outer string
                     data = parser.unescapestr(program[s:pos])
                     if token == b'template':
@@ -300,7 +307,14 b' def _scantemplate(tmpl, start, stop, quo'
                 return
 
             parseres, pos = p.parse(tokenize(tmpl, n + 1, stop, b'}'))
+
+            # pycompat.bytestr (and bytes) both have .startswith() that
+            # takes an optional start and an optional end, but pytype thinks
+            # it only takes 2 args.
+
+            # pytype: disable=wrong-arg-count
             if not tmpl.startswith(b'}', pos):
+                # pytype: enable=wrong-arg-count
                 raise error.ParseError(_(b"invalid token"), pos)
             yield (b'template', parseres, n)
             pos += 1

mercurial/thirdparty/attr/LICENSE

@@ -1,6 +1,6 b''
 The MIT License (MIT)
 
-Copyright (c) 2015 Hynek Schlawack
+Copyright (c) 2015 Hynek Schlawack and the attrs contributors
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

mercurial/thirdparty/attr/__init__.py

@@ -1,37 +1,35 b''
-from __future__ import absolute_import, division, print_function
+# SPDX-License-Identifier: MIT
+
+
+import sys
+
+from functools import partial
 
-from ._funcs import (
-    asdict,
-    assoc,
-    astuple,
-    evolve,
-    has,
-)
+from . import converters, exceptions, filters, setters, validators
+from ._cmp import cmp_using
+from ._config import get_run_validators, set_run_validators
+from ._funcs import asdict, assoc, astuple, evolve, has, resolve_types
 from ._make import (
+    NOTHING,
     Attribute,
     Factory,
-    NOTHING,
-    attr,
-    attributes,
+    attrib,
+    attrs,
     fields,
+    fields_dict,
     make_class,
     validate,
 )
-from ._config import (
-    get_run_validators,
-    set_run_validators,
-)
-from . import exceptions
-from . import filters
-from . import converters
-from . import validators
+from ._version_info import VersionInfo
 
 
-__version__ = "17.2.0"
+__version__ = "22.1.0"
+__version_info__ = VersionInfo._from_version_string(__version__)
 
 __title__ = "attrs"
 __description__ = "Classes Without Boilerplate"
-__uri__ = "http://www.attrs.org/"
+__url__ = "https://www.attrs.org/"
+__uri__ = __url__
 __doc__ = __description__ + " <" + __uri__ + ">"
 
 __author__ = "Hynek Schlawack"
@@ -41,8 +39,9 b' from . import validators'
 __copyright__ = "Copyright (c) 2015 Hynek Schlawack"
 
 
-s = attrs = attributes
-ib = attrib = attr
+s = attributes = attrs
+ib = attr = attrib
+dataclass = partial(attrs, auto_attribs=True)  # happy Easter ;)
 
 __all__ = [
     "Attribute",
@@ -55,17 +54,26 b' ib = attrib = attr'
     "attrib",
     "attributes",
     "attrs",
+    "cmp_using",
     "converters",
     "evolve",
     "exceptions",
     "fields",
+    "fields_dict",
     "filters",
     "get_run_validators",
     "has",
     "ib",
     "make_class",
+    "resolve_types",
     "s",
     "set_run_validators",
+    "setters",
     "validate",
     "validators",
 ]
+
+if sys.version_info[:2] >= (3, 6):
+    from ._next_gen import define, field, frozen, mutable  # noqa: F401
+
+    __all__.extend(("define", "field", "frozen", "mutable"))

mercurial/thirdparty/attr/_compat.py

@@ -1,90 +1,185 b''
-from __future__ import absolute_import, division, print_function
+# SPDX-License-Identifier: MIT
+
+
+import inspect
+import platform
+import sys
+import threading
+import types
+import warnings
+
+from collections.abc import Mapping, Sequence  # noqa
+
+
+PYPY = platform.python_implementation() == "PyPy"
+PY36 = sys.version_info[:2] >= (3, 6)
+HAS_F_STRINGS = PY36
+PY310 = sys.version_info[:2] >= (3, 10)
 
-import sys
-import types
+
+if PYPY or PY36:
+    ordered_dict = dict
+else:
+    from collections import OrderedDict
+
+    ordered_dict = OrderedDict
+
+
+def just_warn(*args, **kw):
+    warnings.warn(
+        "Running interpreter doesn't sufficiently support code object "
+        "introspection.  Some features like bare super() or accessing "
+        "__class__ will not work with slotted classes.",
+        RuntimeWarning,
+        stacklevel=2,
+    )
 
 
-PY2 = sys.version_info[0] == 2
+class _AnnotationExtractor:
+    """
+    Extract type annotations from a callable, returning None whenever there
+    is none.
+    """
+
+    __slots__ = ["sig"]
+
+    def __init__(self, callable):
+        try:
+            self.sig = inspect.signature(callable)
+        except (ValueError, TypeError):  # inspect failed
+            self.sig = None
+
+    def get_first_param_type(self):
+        """
+        Return the type annotation of the first argument if it's not empty.
+        """
+        if not self.sig:
+            return None
+
+        params = list(self.sig.parameters.values())
+        if params and params[0].annotation is not inspect.Parameter.empty:
+            return params[0].annotation
+
+        return None
+
+    def get_return_type(self):
+        """
+        Return the return type if it's not empty.
+        """
+        if (
+            self.sig
+            and self.sig.return_annotation is not inspect.Signature.empty
+        ):
+            return self.sig.return_annotation
+
+        return None
 
 
-if PY2:
-    from UserDict import IterableUserDict
-
-    # We 'bundle' isclass instead of using inspect as importing inspect is
-    # fairly expensive (order of 10-15 ms for a modern machine in 2016)
-    def isclass(klass):
-        return isinstance(klass, (type, types.ClassType))
+def make_set_closure_cell():
+    """Return a function of two arguments (cell, value) which sets
+    the value stored in the closure cell `cell` to `value`.
+    """
+    # pypy makes this easy. (It also supports the logic below, but
+    # why not do the easy/fast thing?)
+    if PYPY:
 
-    # TYPE is used in exceptions, repr(int) is different on Python 2 and 3.
-    TYPE = "type"
+        def set_closure_cell(cell, value):
+            cell.__setstate__((value,))
+
+        return set_closure_cell
 
-    def iteritems(d):
-        return d.iteritems()
+    # Otherwise gotta do it the hard way.
 
-    def iterkeys(d):
-        return d.iterkeys()
+    # Create a function that will set its first cellvar to `value`.
+    def set_first_cellvar_to(value):
+        x = value
+        return
 
-    # Python 2 is bereft of a read-only dict proxy, so we make one!
-    class ReadOnlyDict(IterableUserDict):
-        """
-        Best-effort read-only dict wrapper.
-        """
+        # This function will be eliminated as dead code, but
+        # not before its reference to `x` forces `x` to be
+        # represented as a closure cell rather than a local.
+        def force_x_to_be_a_cell():  # pragma: no cover
+            return x
 
-        def __setitem__(self, key, val):
-            # We gently pretend we're a Python 3 mappingproxy.
-            raise TypeError("'mappingproxy' object does not support item "
-                            "assignment")
+    try:
+        # Extract the code object and make sure our assumptions about
+        # the closure behavior are correct.
+        co = set_first_cellvar_to.__code__
+        if co.co_cellvars != ("x",) or co.co_freevars != ():
+            raise AssertionError  # pragma: no cover
 
-        def update(self, _):
-            # We gently pretend we're a Python 3 mappingproxy.
-            raise AttributeError("'mappingproxy' object has no attribute "
-                                 "'update'")
+        # Convert this code object to a code object that sets the
+        # function's first _freevar_ (not cellvar) to the argument.
+        if sys.version_info >= (3, 8):
 
-        def __delitem__(self, _):
-            # We gently pretend we're a Python 3 mappingproxy.
-            raise TypeError("'mappingproxy' object does not support item "
-                            "deletion")
+            def set_closure_cell(cell, value):
+                cell.cell_contents = value
 
-        def clear(self):
-            # We gently pretend we're a Python 3 mappingproxy.
-            raise AttributeError("'mappingproxy' object has no attribute "
-                                 "'clear'")
-
-        def pop(self, key, default=None):
-            # We gently pretend we're a Python 3 mappingproxy.
-            raise AttributeError("'mappingproxy' object has no attribute "
-                                 "'pop'")
+        else:
+            args = [co.co_argcount]
+            args.append(co.co_kwonlyargcount)
+            args.extend(
+                [
+                    co.co_nlocals,
+                    co.co_stacksize,
+                    co.co_flags,
+                    co.co_code,
+                    co.co_consts,
+                    co.co_names,
+                    co.co_varnames,
+                    co.co_filename,
+                    co.co_name,
+                    co.co_firstlineno,
+                    co.co_lnotab,
+                    # These two arguments are reversed:
+                    co.co_cellvars,
+                    co.co_freevars,
+                ]
+            )
+            set_first_freevar_code = types.CodeType(*args)
 
-        def popitem(self):
-            # We gently pretend we're a Python 3 mappingproxy.
-            raise AttributeError("'mappingproxy' object has no attribute "
-                                 "'popitem'")
-
-        def setdefault(self, key, default=None):
-            # We gently pretend we're a Python 3 mappingproxy.
-            raise AttributeError("'mappingproxy' object has no attribute "
-                                 "'setdefault'")
+            def set_closure_cell(cell, value):
+                # Create a function using the set_first_freevar_code,
+                # whose first closure cell is `cell`. Calling it will
+                # change the value of that cell.
+                setter = types.FunctionType(
+                    set_first_freevar_code, {}, "setter", (), (cell,)
+                )
+                # And call it to set the cell.
+                setter(value)
 
-        def __repr__(self):
-            # Override to be identical to the Python 3 version.
-            return "mappingproxy(" + repr(self.data) + ")"
+        # Make sure it works on this interpreter:
+        def make_func_with_cell():
+            x = None
+
+            def func():
+                return x  # pragma: no cover
 
-    def metadata_proxy(d):
-        res = ReadOnlyDict()
-        res.data.update(d)  # We blocked update, so we have to do it like this.
-        return res
+            return func
+
+        cell = make_func_with_cell().__closure__[0]
+        set_closure_cell(cell, 100)
+        if cell.cell_contents != 100:
+            raise AssertionError  # pragma: no cover
 
-else:
-    def isclass(klass):
-        return isinstance(klass, type)
+    except Exception:
+        return just_warn
+    else:
+        return set_closure_cell
 
-    TYPE = "class"
+
+set_closure_cell = make_set_closure_cell()
 
-    def iteritems(d):
-        return d.items()
-
-    def iterkeys(d):
-        return d.keys()
-
-    def metadata_proxy(d):
-        return types.MappingProxyType(dict(d))
+# Thread-local global to track attrs instances which are already being repr'd.
+# This is needed because there is no other (thread-safe) way to pass info
+# about the instances that are already being repr'd through the call stack
+# in order to ensure we don't perform infinite recursion.
+#
+# For instance, if an instance contains a dict which contains that instance,
+# we need to know that we're already repr'ing the outside instance from within
+# the dict's repr() call.
+#
+# This lives here rather than in _make.py so that the functions in _make.py
+# don't have a direct reference to the thread-local in their globals dict.
+# If they have such a reference, it breaks cloudpickle.
+repr_context = threading.local()

mercurial/thirdparty/attr/_config.py

@@ -1,4 +1,4 b''
-from __future__ import absolute_import, division, print_function
+# SPDX-License-Identifier: MIT
 
 
 __all__ = ["set_run_validators", "get_run_validators"]
@@ -9,6 +9,10 b' from __future__ import absolute_import, '
 def set_run_validators(run):
     """
     Set whether or not validators are run.  By default, they are run.
+
+    .. deprecated:: 21.3.0 It will not be removed, but it also will not be
+        moved to new ``attrs`` namespace. Use `attrs.validators.set_disabled()`
+        instead.
     """
     if not isinstance(run, bool):
         raise TypeError("'run' must be bool.")
@@ -19,5 +23,9 b' def set_run_validators(run):'
 def get_run_validators():
     """
     Return whether or not validators are run.
+
+    .. deprecated:: 21.3.0 It will not be removed, but it also will not be
+        moved to new ``attrs`` namespace. Use `attrs.validators.get_disabled()`
+        instead.
     """
     return _run_validators

mercurial/thirdparty/attr/_funcs.py

@@ -1,14 +1,20 b''
-from __future__ import absolute_import, division, print_function
+# SPDX-License-Identifier: MIT
+
 
 import copy
 
-from ._compat import iteritems
-from ._make import NOTHING, fields, _obj_setattr
+from ._make import NOTHING, _obj_setattr, fields
 from .exceptions import AttrsAttributeNotFoundError
 
 
-def asdict(inst, recurse=True, filter=None, dict_factory=dict,
-           retain_collection_types=False):
+def asdict(
+    inst,
+    recurse=True,
+    filter=None,
+    dict_factory=dict,
+    retain_collection_types=False,
+    value_serializer=None,
+):
     """
     Return the ``attrs`` attribute values of *inst* as a dict.
 
@@ -17,9 +23,9 b' def asdict(inst, recurse=True, filter=No'
     :param inst: Instance of an ``attrs``-decorated class.
     :param bool recurse: Recurse into classes that are also
         ``attrs``-decorated.
-    :param callable filter: A callable whose return code deteremines whether an
+    :param callable filter: A callable whose return code determines whether an
         attribute or element is included (``True``) or dropped (``False``).  Is
-        called with the :class:`attr.Attribute` as the first argument and the
+        called with the `attrs.Attribute` as the first argument and the
         value as the second argument.
     :param callable dict_factory: A callable to produce dictionaries from.  For
         example, to produce ordered dictionaries instead of normal Python
@@ -27,6 +33,10 b' def asdict(inst, recurse=True, filter=No'
     :param bool retain_collection_types: Do not convert to ``list`` when
         encountering an attribute whose type is ``tuple`` or ``set``.  Only
         meaningful if ``recurse`` is ``True``.
+    :param Optional[callable] value_serializer: A hook that is called for every
+        attribute or dict key/value.  It receives the current instance, field
+        and value and must return the (updated) value.  The hook is run *after*
+        the optional *filter* has been applied.
 
     :rtype: return type of *dict_factory*
 
@@ -35,6 +45,9 b' def asdict(inst, recurse=True, filter=No'
 
     ..  versionadded:: 16.0.0 *dict_factory*
     ..  versionadded:: 16.1.0 *retain_collection_types*
+    ..  versionadded:: 20.3.0 *value_serializer*
+    ..  versionadded:: 21.3.0 If a dict has a collection for a key, it is
+        serialized as a tuple.
     """
     attrs = fields(inst.__class__)
     rv = dict_factory()
@@ -42,24 +55,58 b' def asdict(inst, recurse=True, filter=No'
         v = getattr(inst, a.name)
         if filter is not None and not filter(a, v):
             continue
+
+        if value_serializer is not None:
+            v = value_serializer(inst, a, v)
+
         if recurse is True:
             if has(v.__class__):
-                rv[a.name] = asdict(v, recurse=True, filter=filter,
-                                    dict_factory=dict_factory)
-            elif isinstance(v, (tuple, list, set)):
+                rv[a.name] = asdict(
+                    v,
+                    recurse=True,
+                    filter=filter,
+                    dict_factory=dict_factory,
+                    retain_collection_types=retain_collection_types,
+                    value_serializer=value_serializer,
+                )
+            elif isinstance(v, (tuple, list, set, frozenset)):
                 cf = v.__class__ if retain_collection_types is True else list
-                rv[a.name] = cf([
-                    asdict(i, recurse=True, filter=filter,
-                           dict_factory=dict_factory)
-                    if has(i.__class__) else i
-                    for i in v
-                ])
+                rv[a.name] = cf(
+                    [
+                        _asdict_anything(
+                            i,
+                            is_key=False,
+                            filter=filter,
+                            dict_factory=dict_factory,
+                            retain_collection_types=retain_collection_types,
+                            value_serializer=value_serializer,
+                        )
+                        for i in v
+                    ]
+                )
             elif isinstance(v, dict):
                 df = dict_factory
-                rv[a.name] = df((
-                    asdict(kk, dict_factory=df) if has(kk.__class__) else kk,
-                    asdict(vv, dict_factory=df) if has(vv.__class__) else vv)
-                    for kk, vv in iteritems(v))
+                rv[a.name] = df(
+                    (
+                        _asdict_anything(
+                            kk,
+                            is_key=True,
+                            filter=filter,
+                            dict_factory=df,
+                            retain_collection_types=retain_collection_types,
+                            value_serializer=value_serializer,
+                        ),
+                        _asdict_anything(
+                            vv,
+                            is_key=False,
+                            filter=filter,
+                            dict_factory=df,
+                            retain_collection_types=retain_collection_types,
+                            value_serializer=value_serializer,
+                        ),
+                    )
+                    for kk, vv in v.items()
+                )
             else:
                 rv[a.name] = v
         else:
@@ -67,8 +114,86 b' def asdict(inst, recurse=True, filter=No'
     return rv
 
 
-def astuple(inst, recurse=True, filter=None, tuple_factory=tuple,
-            retain_collection_types=False):
+def _asdict_anything(
+    val,
+    is_key,
+    filter,
+    dict_factory,
+    retain_collection_types,
+    value_serializer,
+):
+    """
+    ``asdict`` only works on attrs instances, this works on anything.
+    """
+    if getattr(val.__class__, "__attrs_attrs__", None) is not None:
+        # Attrs class.
+        rv = asdict(
+            val,
+            recurse=True,
+            filter=filter,
+            dict_factory=dict_factory,
+            retain_collection_types=retain_collection_types,
+            value_serializer=value_serializer,
+        )
+    elif isinstance(val, (tuple, list, set, frozenset)):
+        if retain_collection_types is True:
+            cf = val.__class__
+        elif is_key:
+            cf = tuple
+        else:
+            cf = list
+
+        rv = cf(
+            [
+                _asdict_anything(
+                    i,
+                    is_key=False,
+                    filter=filter,
+                    dict_factory=dict_factory,
+                    retain_collection_types=retain_collection_types,
+                    value_serializer=value_serializer,
+                )
+                for i in val
+            ]
+        )
+    elif isinstance(val, dict):
+        df = dict_factory
+        rv = df(
+            (
+                _asdict_anything(
+                    kk,
+                    is_key=True,
+                    filter=filter,
+                    dict_factory=df,
+                    retain_collection_types=retain_collection_types,
+                    value_serializer=value_serializer,
+                ),
+                _asdict_anything(
+                    vv,
+                    is_key=False,
+                    filter=filter,
+                    dict_factory=df,
+                    retain_collection_types=retain_collection_types,
+                    value_serializer=value_serializer,
+                ),
+            )
+            for kk, vv in val.items()
+        )
+    else:
+        rv = val
+        if value_serializer is not None:
+            rv = value_serializer(None, None, rv)
+
+    return rv
+
+
+def astuple(
+    inst,
+    recurse=True,
+    filter=None,
+    tuple_factory=tuple,
+    retain_collection_types=False,
+):
     """
     Return the ``attrs`` attribute values of *inst* as a tuple.
 
@@ -79,7 +204,7 b' def astuple(inst, recurse=True, filter=N'
         ``attrs``-decorated.
     :param callable filter: A callable whose return code determines whether an
         attribute or element is included (``True``) or dropped (``False``).  Is
-        called with the :class:`attr.Attribute` as the first argument and the
+        called with the `attrs.Attribute` as the first argument and the
         value as the second argument.
     :param callable tuple_factory: A callable to produce tuples from.  For
         example, to produce lists instead of tuples.
@@ -104,38 +229,61 b' def astuple(inst, recurse=True, filter=N'
             continue
         if recurse is True:
             if has(v.__class__):
-                rv.append(astuple(v, recurse=True, filter=filter,
-                                  tuple_factory=tuple_factory,
-                                  retain_collection_types=retain))
-            elif isinstance(v, (tuple, list, set)):
+                rv.append(
+                    astuple(
+                        v,
+                        recurse=True,
+                        filter=filter,
+                        tuple_factory=tuple_factory,
+                        retain_collection_types=retain,
+                    )
+                )
+            elif isinstance(v, (tuple, list, set, frozenset)):
                 cf = v.__class__ if retain is True else list
-                rv.append(cf([
-                    astuple(j, recurse=True, filter=filter,
-                            tuple_factory=tuple_factory,
-                            retain_collection_types=retain)
-                    if has(j.__class__) else j
-                    for j in v
-                ]))
+                rv.append(
+                    cf(
+                        [
+                            astuple(
+                                j,
+                                recurse=True,
+                                filter=filter,
+                                tuple_factory=tuple_factory,
+                                retain_collection_types=retain,
+                            )
+                            if has(j.__class__)
+                            else j
+                            for j in v
+                        ]
+                    )
+                )
             elif isinstance(v, dict):
                 df = v.__class__ if retain is True else dict
-                rv.append(df(
+                rv.append(
+                    df(
                         (
                             astuple(
                                 kk,
                                 tuple_factory=tuple_factory,
-                                retain_collection_types=retain
-                            ) if has(kk.__class__) else kk,
+                                retain_collection_types=retain,
+                            )
+                            if has(kk.__class__)
+                            else kk,
                             astuple(
                                 vv,
                                 tuple_factory=tuple_factory,
-                                retain_collection_types=retain
-                            ) if has(vv.__class__) else vv
+                                retain_collection_types=retain,
+                            )
+                            if has(vv.__class__)
+                            else vv,
                         )
-                        for kk, vv in iteritems(v)))
+                        for kk, vv in v.items()
+                    )
+                )
             else:
                 rv.append(v)
         else:
             rv.append(v)
+
     return rv if tuple_factory is list else tuple_factory(rv)
 
 
@@ -146,7 +294,7 b' def has(cls):'
     :param type cls: Class to introspect.
     :raise TypeError: If *cls* is not a class.
 
-    :rtype: :class:`bool`
+    :rtype: bool
     """
     return getattr(cls, "__attrs_attrs__", None) is not None
 
@@ -166,19 +314,26 b' def assoc(inst, **changes):'
         class.
 
     ..  deprecated:: 17.1.0
-        Use :func:`evolve` instead.
+        Use `attrs.evolve` instead if you can.
+        This function will not be removed du to the slightly different approach
+        compared to `attrs.evolve`.
     """
     import warnings
-    warnings.warn("assoc is deprecated and will be removed after 2018/01.",
-                  DeprecationWarning)
+
+    warnings.warn(
+        "assoc is deprecated and will be removed after 2018/01.",
+        DeprecationWarning,
+        stacklevel=2,
+    )
     new = copy.copy(inst)
     attrs = fields(inst.__class__)
-    for k, v in iteritems(changes):
+    for k, v in changes.items():
         a = getattr(attrs, k, NOTHING)
         if a is NOTHING:
             raise AttrsAttributeNotFoundError(
-                "{k} is not an attrs attribute on {cl}."
-                .format(k=k, cl=new.__class__)
+                "{k} is not an attrs attribute on {cl}.".format(
+                    k=k, cl=new.__class__
+                )
             )
         _obj_setattr(new, k, v)
     return new
@@ -209,4 +364,57 b' def evolve(inst, **changes):'
         init_name = attr_name if attr_name[0] != "_" else attr_name[1:]
         if init_name not in changes:
             changes[init_name] = getattr(inst, attr_name)
+
     return cls(**changes)
+
+
+def resolve_types(cls, globalns=None, localns=None, attribs=None):
+    """
+    Resolve any strings and forward annotations in type annotations.
+
+    This is only required if you need concrete types in `Attribute`'s *type*
+    field. In other words, you don't need to resolve your types if you only
+    use them for static type checking.
+
+    With no arguments, names will be looked up in the module in which the class
+    was created. If this is not what you want, e.g. if the name only exists
+    inside a method, you may pass *globalns* or *localns* to specify other
+    dictionaries in which to look up these names. See the docs of
+    `typing.get_type_hints` for more details.
+
+    :param type cls: Class to resolve.
+    :param Optional[dict] globalns: Dictionary containing global variables.
+    :param Optional[dict] localns: Dictionary containing local variables.
+    :param Optional[list] attribs: List of attribs for the given class.
+        This is necessary when calling from inside a ``field_transformer``
+        since *cls* is not an ``attrs`` class yet.
+
+    :raise TypeError: If *cls* is not a class.
+    :raise attr.exceptions.NotAnAttrsClassError: If *cls* is not an ``attrs``
+        class and you didn't pass any attribs.
+    :raise NameError: If types cannot be resolved because of missing variables.
+
+    :returns: *cls* so you can use this function also as a class decorator.
+        Please note that you have to apply it **after** `attrs.define`. That
+        means the decorator has to come in the line **before** `attrs.define`.
+
+    ..  versionadded:: 20.1.0
+    ..  versionadded:: 21.1.0 *attribs*
+
+    """
+    # Since calling get_type_hints is expensive we cache whether we've
+    # done it already.
+    if getattr(cls, "__attrs_types_resolved__", None) != cls:
+        import typing
+
+        hints = typing.get_type_hints(cls, globalns=globalns, localns=localns)
+        for field in fields(cls) if attribs is None else attribs:
+            if field.name in hints:
+                # Since fields have been frozen we must work around it.
+                _obj_setattr(field, "type", hints[field.name])
+        # We store the class we resolved so that subclasses know they haven't
+        # been resolved.
+        cls.__attrs_types_resolved__ = cls
+
+    # Return the class so you can use it as a decorator too.
+    return cls