dirstate: invalidate on all exceptions

Previously, we would miss SystemExit, KeyboardInterrupt etc.

This "fix" on the bug tested in "test-largefiles-update.t" by preventing the
precisely tested situation to happens at all. However this reveal a similar bug
with a different timing.

I have not been able to deal with that pre-existing bug so far. So I updated the
test to point that out.

mercurial/dirstate.py

@@ -1,1660 +1,1660 b''
 # dirstate.py - working directory tracking for mercurial
 #
 # Copyright 2005-2007 Olivia Mackall <[e-mail]>
 #
 # 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 contextlib
 import os
 import stat
 import uuid
 
 from .i18n import _
 from .pycompat import delattr
 
 from hgdemandimport import tracing
 
 from . import (
     dirstatemap,
     encoding,
     error,
     match as matchmod,
     node,
     pathutil,
     policy,
     pycompat,
     scmutil,
     util,
 )
 
 from .dirstateutils import (
     timestamp,
 )
 
 from .interfaces import (
     dirstate as intdirstate,
     util as interfaceutil,
 )
 
 parsers = policy.importmod('parsers')
 rustmod = policy.importrust('dirstate')
 
 HAS_FAST_DIRSTATE_V2 = rustmod is not None
 
 propertycache = util.propertycache
 filecache = scmutil.filecache
 _rangemask = dirstatemap.rangemask
 
 DirstateItem = dirstatemap.DirstateItem
 
 
 class repocache(filecache):
     """filecache for files in .hg/"""
 
     def join(self, obj, fname):
         return obj._opener.join(fname)
 
 
 class rootcache(filecache):
     """filecache for files in the repository root"""
 
     def join(self, obj, fname):
         return obj._join(fname)
 
 
 def requires_changing_parents(func):
     def wrap(self, *args, **kwargs):
         if not self.is_changing_parents:
             msg = 'calling `%s` outside of a changing_parents context'
             msg %= func.__name__
             raise error.ProgrammingError(msg)
         if self._invalidated_context:
             msg = 'calling `%s` after the dirstate was invalidated'
             raise error.ProgrammingError(msg)
         return func(self, *args, **kwargs)
 
     return 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
 
 
 def requires_not_changing_parents(func):
     def wrap(self, *args, **kwargs):
         if self.is_changing_parents:
             msg = 'calling `%s` inside of a changing_parents context'
             msg %= func.__name__
             raise error.ProgrammingError(msg)
         return func(self, *args, **kwargs)
 
     return wrap
 
 
 CHANGE_TYPE_PARENTS = "parents"
 CHANGE_TYPE_FILES = "files"
 
 
 @interfaceutil.implementer(intdirstate.idirstate)
 class dirstate:
     def __init__(
         self,
         opener,
         ui,
         root,
         validate,
         sparsematchfn,
         nodeconstants,
         use_dirstate_v2,
         use_tracked_hint=False,
     ):
         """Create a new dirstate object.
 
         opener is an open()-like callable that can be used to open the
         dirstate file; root is the root of the directory tracked by
         the dirstate.
         """
         self._use_dirstate_v2 = use_dirstate_v2
         self._use_tracked_hint = use_tracked_hint
         self._nodeconstants = nodeconstants
         self._opener = opener
         self._validate = validate
         self._root = root
         # Either build a sparse-matcher or None if sparse is disabled
         self._sparsematchfn = sparsematchfn
         # ntpath.join(root, '') of Python 2.7.9 does not add sep if root is
         # UNC path pointing to root share (issue4557)
         self._rootdir = pathutil.normasprefix(root)
         # True is any internal state may be different
         self._dirty = False
         # True if the set of tracked file may be different
         self._dirty_tracked_set = False
         self._ui = ui
         self._filecache = {}
         # nesting level of `changing_parents` context
         self._changing_level = 0
         # the change currently underway
         self._change_type = None
         # 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._filename = b'dirstate'
         self._filename_th = b'dirstate-tracked-hint'
         self._pendingfilename = b'%s.pending' % self._filename
         self._plchangecallbacks = {}
         self._origpl = None
         self._mapcls = dirstatemap.dirstatemap
         # Access and cache cwd early, so we don't access it for the first time
         # after a working-copy update caused it to not exist (accessing it then
         # raises an exception).
         self._cwd
 
     def prefetch_parents(self):
         """make sure the parents are loaded
 
         Used to avoid a race condition.
         """
         self._pl
 
     @contextlib.contextmanager
     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)
         if self._invalidated_context:
             msg = "trying to use an invalidated dirstate before it has reset"
             raise error.ProgrammingError(msg)
 
         has_tr = repo.currenttransaction() is not None
 
         # 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)
         self._changing_level += 1
         try:
             yield
-        except Exception:
+        except:  # re-raises
             self.invalidate()
             raise
         finally:
             tr = repo.currenttransaction()
             if self._changing_level > 0:
                 if self._invalidated_context:
                     # make sure we invalidate anything an upper context might
                     # have changed.
                     self.invalidate()
                 self._changing_level -= 1
                 # The invalidation is complete once we exit the final context
                 # manager
                 if self._changing_level <= 0:
                     self._change_type = None
                     assert self._changing_level == 0
                     if self._invalidated_context:
                         self._invalidated_context = False
                     else:
                         # When an exception occured, `_invalidated_context`
                         # would have been set to True by the `invalidate`
                         # call earlier.
                         #
                         # We don't have more straightforward code, because the
                         # Exception catching (and the associated `invalidate`
                         # calling) might have been called by a nested context
                         # instead of the top level one.
                         self.write(tr)
             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)
 
     @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.
         """
         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(
             self._ui,
             self._opener,
             self._root,
             self._nodeconstants,
             self._use_dirstate_v2,
         )
         return self._map
 
     @property
     def _sparsematcher(self):
         """The matcher for the sparse checkout.
 
         The working directory may not include every file from a manifest. The
         matcher obtained by this property will match a path if it is to be
         included in the working directory.
 
         When sparse if disabled, return None.
         """
         if self._sparsematchfn is None:
             return None
         # TODO there is potential to cache this property. For now, the matcher
         # is resolved on every access. (But the called function does use a
         # cache to keep the lookup fast.)
         return self._sparsematchfn()
 
     @repocache(b'branch')
     def _branch(self):
         try:
             return self._opener.read(b"branch").strip() or b"default"
         except FileNotFoundError:
             return b"default"
 
     @property
     def _pl(self):
         return self._map.parents()
 
     def hasdir(self, d):
         return self._map.hastrackeddir(d)
 
     @rootcache(b'.hgignore')
     def _ignore(self):
         files = self._ignorefiles()
         if not files:
             return matchmod.never()
 
         pats = [b'include:%s' % f for f in files]
         return matchmod.match(self._root, b'', [], pats, warn=self._ui.warn)
 
     @propertycache
     def _slash(self):
         return self._ui.configbool(b'ui', b'slash') and pycompat.ossep != b'/'
 
     @propertycache
     def _checklink(self):
         return util.checklink(self._root)
 
     @propertycache
     def _checkexec(self):
         return bool(util.checkexec(self._root))
 
     @propertycache
     def _checkcase(self):
         return not util.fscasesensitive(self._join(b'.hg'))
 
     def _join(self, f):
         # much faster than os.path.join()
         # it's safe because f is always a relative path
         return self._rootdir + f
 
     def flagfunc(self, buildfallback):
         """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.
         """
 
         # small hack to cache the result of buildfallback()
         fallback_func = []
 
         def get_flags(x):
             entry = None
             fallback_value = None
             try:
                 st = os.lstat(self._join(x))
             except OSError:
                 return b''
 
             if self._checklink:
                 if util.statislink(st):
                     return b'l'
             else:
                 entry = self.get_entry(x)
                 if entry.has_fallback_symlink:
                     if entry.fallback_symlink:
                         return b'l'
                 else:
                     if not fallback_func:
                         fallback_func.append(buildfallback())
                     fallback_value = fallback_func[0](x)
                     if b'l' in fallback_value:
                         return b'l'
 
             if self._checkexec:
                 if util.statisexec(st):
                     return b'x'
             else:
                 if entry is None:
                     entry = self.get_entry(x)
                 if entry.has_fallback_exec:
                     if entry.fallback_exec:
                         return b'x'
                 else:
                     if fallback_value is None:
                         if not fallback_func:
                             fallback_func.append(buildfallback())
                         fallback_value = fallback_func[0](x)
                     if b'x' in fallback_value:
                         return b'x'
             return b''
 
         return get_flags
 
     @propertycache
     def _cwd(self):
         # internal config: ui.forcecwd
         forcecwd = self._ui.config(b'ui', b'forcecwd')
         if forcecwd:
             return forcecwd
         return encoding.getcwd()
 
     def getcwd(self):
         """Return the path from which a canonical path is calculated.
 
         This path should be used to resolve file patterns or to convert
         canonical paths back to file paths for display. It shouldn't be
         used to get real file paths. Use vfs functions instead.
         """
         cwd = self._cwd
         if cwd == self._root:
             return b''
         # self._root ends with a path separator if self._root is '/' or 'C:\'
         rootsep = self._root
         if not util.endswithsep(rootsep):
             rootsep += pycompat.ossep
         if cwd.startswith(rootsep):
             return cwd[len(rootsep) :]
         else:
             # we're outside the repo. return an absolute path.
             return cwd
 
     def pathto(self, f, cwd=None):
         if cwd is None:
             cwd = self.getcwd()
         path = util.pathto(self._root, cwd, f)
         if self._slash:
             return util.pconvert(path)
         return path
 
     def get_entry(self, path):
         """return a DirstateItem for the associated path"""
         entry = self._map.get(path)
         if entry is None:
             return DirstateItem()
         return entry
 
     def __contains__(self, key):
         return key in self._map
 
     def __iter__(self):
         return iter(sorted(self._map))
 
     def items(self):
         return self._map.items()
 
     iteritems = items
 
     def parents(self):
         return [self._validate(p) for p in self._pl]
 
     def p1(self):
         return self._validate(self._pl[0])
 
     def p2(self):
         return self._validate(self._pl[1])
 
     @property
     def in_merge(self):
         """True if a merge is in progress"""
         return self._pl[1] != self._nodeconstants.nullid
 
     def branch(self):
         return encoding.tolocal(self._branch)
 
     # XXX since this make the dirstate dirty, we should enforce that it is done
     # withing an appropriate change-context that scope the change and ensure it
     # eventually get written on disk (or rolled back)
     def setparents(self, p1, p2=None):
         """Set dirstate parents to p1 and p2.
 
         When moving from two parents to one, "merged" entries a
         adjusted to normal and previous copy records discarded and
         returned by the call.
 
         See localrepo.setparents()
         """
         if p2 is None:
             p2 = self._nodeconstants.nullid
         if self._changing_level == 0:
             raise ValueError(
                 b"cannot set dirstate parent outside of "
                 b"dirstate.changing_parents context manager"
             )
 
         self._dirty = True
         oldp2 = self._pl[1]
         if self._origpl is None:
             self._origpl = self._pl
         nullid = self._nodeconstants.nullid
         # True if we need to fold p2 related state back to a linear case
         fold_p2 = oldp2 != nullid and p2 == nullid
         return self._map.setparents(p1, p2, fold_p2=fold_p2)
 
     def setbranch(self, branch):
         self.__class__._branch.set(self, encoding.fromlocal(branch))
         f = self._opener(b'branch', b'w', atomictemp=True, checkambig=True)
         try:
             f.write(self._branch + b'\n')
             f.close()
 
             # make sure filecache has the correct stat info for _branch after
             # replacing the underlying file
             ce = self._filecache[b'_branch']
             if ce:
                 ce.refresh()
         except:  # re-raises
             f.discard()
             raise
 
     def invalidate(self):
         """Causes the next access to reread the dirstate.
 
         This is different from localrepo.invalidatedirstate() because it always
         rereads the dirstate. Use localrepo.invalidatedirstate() if you want to
         check whether the dirstate has changed before rereading it."""
 
         for a in ("_map", "_branch", "_ignore"):
             if a in self.__dict__:
                 delattr(self, a)
         self._dirty = False
         self._dirty_tracked_set = False
         self._invalidated_context = self._changing_level > 0
         self._origpl = None
 
     # XXX since this make the dirstate dirty, we should enforce that it is done
     # withing an appropriate change-context that scope the change and ensure it
     # eventually get written on disk (or rolled back)
     def copy(self, source, dest):
         """Mark dest as a copy of source. Unmark dest if source is None."""
         if source == dest:
             return
         self._dirty = True
         if source is not None:
             self._check_sparse(source)
             self._map.copymap[dest] = source
         else:
             self._map.copymap.pop(dest, None)
 
     def copied(self, file):
         return self._map.copymap.get(file, None)
 
     def copies(self):
         return self._map.copymap
 
     @requires_changing_files
     def set_tracked(self, filename, reset_copy=False):
         """a "public" method for generic code to mark a file as tracked
 
         This function is to be called outside of "update/merge" case. For
         example by a command like `hg add X`.
 
         if reset_copy is set, any existing copy information will be dropped.
 
         return True the file was previously untracked, False otherwise.
         """
         self._dirty = True
         entry = self._map.get(filename)
         if entry is None or not entry.tracked:
             self._check_new_tracked_filename(filename)
         pre_tracked = self._map.set_tracked(filename)
         if reset_copy:
             self._map.copymap.pop(filename, None)
         if pre_tracked:
             self._dirty_tracked_set = True
         return pre_tracked
 
     @requires_changing_files
     def set_untracked(self, filename):
         """a "public" method for generic code to mark a file as untracked
 
         This function is to be called outside of "update/merge" case. For
         example by a command like `hg remove X`.
 
         return True the file was previously tracked, False otherwise.
         """
         ret = self._map.set_untracked(filename)
         if ret:
             self._dirty = True
             self._dirty_tracked_set = True
         return ret
 
     @requires_not_changing_parents
     def set_clean(self, filename, parentfiledata):
         """record that the current state of the file on disk is known to be clean"""
         self._dirty = True
         if not self._map[filename].tracked:
             self._check_new_tracked_filename(filename)
         (mode, size, mtime) = parentfiledata
         self._map.set_clean(filename, mode, size, mtime)
 
     @requires_not_changing_parents
     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_changing_parents
     def update_file_p1(
         self,
         filename,
         p1_tracked,
     ):
         """Set a file as tracked in the parent (or not)
 
         This is to be called when adjust the dirstate to a new parent after an history
         rewriting operation.
 
         It should not be called during a merge (p2 != nullid) and only within
         a `with dirstate.changing_parents(repo):` context.
         """
         if self.in_merge:
             msg = b'update_file_reference should not be called when merging'
             raise error.ProgrammingError(msg)
         entry = self._map.get(filename)
         if entry is None:
             wc_tracked = False
         else:
             wc_tracked = entry.tracked
         if not (p1_tracked or wc_tracked):
             # the file is no longer relevant to anyone
             if self._map.get(filename) is not None:
                 self._map.reset_state(filename)
                 self._dirty = True
         elif (not p1_tracked) and wc_tracked:
             if entry is not None and entry.added:
                 return  # avoid dropping copy information (maybe?)
 
         self._map.reset_state(
             filename,
             wc_tracked,
             p1_tracked,
             # the underlying reference might have changed, we will have to
             # check it.
             has_meaningful_mtime=False,
         )
 
     @requires_changing_parents
     def update_file(
         self,
         filename,
         wc_tracked,
         p1_tracked,
         p2_info=False,
         possibly_dirty=False,
         parentfiledata=None,
     ):
         """update the information about a file in the 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.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,
         )
 
     # XXX since this make the dirstate dirty, we should enforce that it is done
     # withing an appropriate change-context that scope the change and ensure it
     # eventually get written on disk (or rolled back)
     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.
         """
         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
         # this. The test agrees
 
         self._dirty = True
         old_entry = self._map.get(filename)
         if old_entry is None:
             prev_tracked = False
         else:
             prev_tracked = old_entry.tracked
         if prev_tracked != wc_tracked:
             self._dirty_tracked_set = True
 
         self._map.reset_state(
             filename,
             wc_tracked,
             p1_tracked,
             p2_info=p2_info,
             has_meaningful_mtime=not possibly_dirty,
             parentfiledata=parentfiledata,
         )
 
     def _check_new_tracked_filename(self, filename):
         scmutil.checkfilename(filename)
         if self._map.hastrackeddir(filename):
             msg = _(b'directory %r already in dirstate')
             msg %= pycompat.bytestr(filename)
             raise error.Abort(msg)
         # shadows
         for d in pathutil.finddirs(filename):
             if self._map.hastrackeddir(d):
                 break
             entry = self._map.get(d)
             if entry is not None and not entry.removed:
                 msg = _(b'file %r in dirstate clashes with %r')
                 msg %= (pycompat.bytestr(d), pycompat.bytestr(filename))
                 raise error.Abort(msg)
         self._check_sparse(filename)
 
     def _check_sparse(self, filename):
         """Check that a filename is inside the sparse profile"""
         sparsematch = self._sparsematcher
         if sparsematch is not None and not sparsematch.always():
             if not sparsematch(filename):
                 msg = _(b"cannot add '%s' - it is outside the sparse checkout")
                 hint = _(
                     b'include file with `hg debugsparse --include <pattern>` or use '
                     b'`hg add -s <file>` to include file directory while adding'
                 )
                 raise error.Abort(msg % filename, hint=hint)
 
     def _discoverpath(self, path, normed, ignoremissing, exists, storemap):
         if exists is None:
             exists = os.path.lexists(os.path.join(self._root, path))
         if not exists:
             # Maybe a path component exists
             if not ignoremissing and b'/' in path:
                 d, f = path.rsplit(b'/', 1)
                 d = self._normalize(d, False, ignoremissing, None)
                 folded = d + b"/" + f
             else:
                 # No path components, preserve original case
                 folded = path
         else:
             # recursively normalize leading directory components
             # against dirstate
             if b'/' in normed:
                 d, f = normed.rsplit(b'/', 1)
                 d = self._normalize(d, False, ignoremissing, True)
                 r = self._root + b"/" + d
                 folded = d + b"/" + util.fspath(f, r)
             else:
                 folded = util.fspath(normed, self._root)
             storemap[normed] = folded
 
         return folded
 
     def _normalizefile(self, path, isknown, ignoremissing=False, exists=None):
         normed = util.normcase(path)
         folded = self._map.filefoldmap.get(normed, None)
         if folded is None:
             if isknown:
                 folded = path
             else:
                 folded = self._discoverpath(
                     path, normed, ignoremissing, exists, self._map.filefoldmap
                 )
         return folded
 
     def _normalize(self, path, isknown, ignoremissing=False, exists=None):
         normed = util.normcase(path)
         folded = self._map.filefoldmap.get(normed, None)
         if folded is None:
             folded = self._map.dirfoldmap.get(normed, None)
         if folded is None:
             if isknown:
                 folded = path
             else:
                 # store discovered result in dirfoldmap so that future
                 # normalizefile calls don't start matching directories
                 folded = self._discoverpath(
                     path, normed, ignoremissing, exists, self._map.dirfoldmap
                 )
         return folded
 
     def normalize(self, path, isknown=False, ignoremissing=False):
         """
         normalize the case of a pathname when on a casefolding filesystem
 
         isknown specifies whether the filename came from walking the
         disk, to avoid extra filesystem access.
 
         If ignoremissing is True, missing path are returned
         unchanged. Otherwise, we try harder to normalize possibly
         existing path components.
 
         The normalized case is determined based on the following precedence:
 
         - version of name already stored in the dirstate
         - version of name stored on disk
         - version provided via command arguments
         """
 
         if self._checkcase:
             return self._normalize(path, isknown, ignoremissing)
         return path
 
     # XXX since this make the dirstate dirty, we should enforce that it is done
     # withing an appropriate change-context that scope the change and ensure it
     # eventually get written on disk (or rolled back)
     def clear(self):
         self._map.clear()
         self._dirty = True
 
     # XXX since this make the dirstate dirty, we should enforce that it is done
     # withing an appropriate change-context that scope the change and ensure it
     # eventually get written on disk (or rolled back)
     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
             allfiles = [f for f in allfiles if matcher(f)]
             if changedfiles:
                 changedfiles = [f for f in changedfiles if matcher(f)]
 
             if changedfiles is not None:
                 # these files will be deleted from the dirstate when they are
                 # not found to be in allfiles
                 dirstatefilestoremove = {f for f in self if not matcher(f)}
                 changedfiles = dirstatefilestoremove.union(changedfiles)
 
         if changedfiles is None:
             # Rebuild entire dirstate
             to_lookup = allfiles
             to_drop = []
             self.clear()
         elif len(changedfiles) < 10:
             # Avoid turning allfiles into a set, which can be expensive if it's
             # large.
             to_lookup = []
             to_drop = []
             for f in changedfiles:
                 if f in allfiles:
                     to_lookup.append(f)
                 else:
                     to_drop.append(f)
         else:
             changedfilesset = set(changedfiles)
             to_lookup = changedfilesset & set(allfiles)
             to_drop = changedfilesset - to_lookup
 
         if self._origpl is None:
             self._origpl = self._pl
         self._map.setparents(parent, self._nodeconstants.nullid)
 
         for f in to_lookup:
             if self.in_merge:
                 self.set_tracked(f)
             else:
                 self._map.reset_state(
                     f,
                     wc_tracked=True,
                     p1_tracked=True,
                 )
         for f in to_drop:
             self._map.reset_state(f)
 
         self._dirty = True
 
     def identity(self):
         """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.
         """
         return self._map.identity
 
     def write(self, tr):
         if not self._dirty:
             return
 
         write_key = self._use_tracked_hint and self._dirty_tracked_set
         if tr:
             # make sure we invalidate the current change on abort
             if tr is not None:
                 tr.addabort(
                     b'dirstate-invalidate',
                     lambda tr: self.invalidate(),
                 )
             # delay writing in-memory changes out
             tr.addfilegenerator(
                 b'dirstate-1-main',
                 (self._filename,),
                 lambda f: self._writedirstate(tr, f),
                 location=b'plain',
                 post_finalize=True,
             )
             if write_key:
                 tr.addfilegenerator(
                     b'dirstate-2-key-post',
                     (self._filename_th,),
                     lambda f: self._write_tracked_hint(tr, f),
                     location=b'plain',
                     post_finalize=True,
                 )
             return
 
         file = lambda f: self._opener(f, b"w", atomictemp=True, checkambig=True)
         with file(self._filename) as f:
             self._writedirstate(tr, f)
         if write_key:
             # we update the key-file after writing to make sure reader have a
             # key that match the newly written content
             with file(self._filename_th) as f:
                 self._write_tracked_hint(tr, f)
 
     def delete_tracked_hint(self):
         """remove the tracked_hint file
 
         To be used by format downgrades operation"""
         self._opener.unlink(self._filename_th)
         self._use_tracked_hint = False
 
     def addparentchangecallback(self, category, callback):
         """add a callback to be called when the wd parents are changed
 
         Callback will be called with the following arguments:
             dirstate, (oldp1, oldp2), (newp1, newp2)
 
         Category is a unique identifier to allow overwriting an old callback
         with a newer callback.
         """
         self._plchangecallbacks[category] = callback
 
     def _writedirstate(self, tr, st):
         # notify callbacks about parents change
         if self._origpl is not None and self._origpl != self._pl:
             for c, callback in sorted(self._plchangecallbacks.items()):
                 callback(self, self._origpl, self._pl)
             self._origpl = None
         self._map.write(tr, st)
         self._dirty = False
         self._dirty_tracked_set = False
 
     def _write_tracked_hint(self, tr, f):
         key = node.hex(uuid.uuid4().bytes)
         f.write(b"1\n%s\n" % key)  # 1 is the format version
 
     def _dirignore(self, f):
         if self._ignore(f):
             return True
         for p in pathutil.finddirs(f):
             if self._ignore(p):
                 return True
         return False
 
     def _ignorefiles(self):
         files = []
         if os.path.exists(self._join(b'.hgignore')):
             files.append(self._join(b'.hgignore'))
         for name, path in self._ui.configitems(b"ui"):
             if name == b'ignore' or name.startswith(b'ignore.'):
                 # we need to use os.path.join here rather than self._join
                 # because path is arbitrary and user-specified
                 files.append(os.path.join(self._rootdir, util.expandpath(path)))
         return files
 
     def _ignorefileandline(self, f):
         files = collections.deque(self._ignorefiles())
         visited = set()
         while files:
             i = files.popleft()
             patterns = matchmod.readpatternfile(
                 i, self._ui.warn, sourceinfo=True
             )
             for pattern, lineno, line in patterns:
                 kind, p = matchmod._patsplit(pattern, b'glob')
                 if kind == b"subinclude":
                     if p not in visited:
                         files.append(p)
                     continue
                 m = matchmod.match(
                     self._root, b'', [], [pattern], warn=self._ui.warn
                 )
                 if m(f):
                     return (i, lineno, line)
             visited.add(i)
         return (None, -1, b"")
 
     def _walkexplicit(self, match, subrepos):
         """Get stat data about the files explicitly specified by match.
 
         Return a triple (results, dirsfound, dirsnotfound).
         - results is a mapping from filename to stat result. It also contains
           listings mapping subrepos and .hg to None.
         - dirsfound is a list of files found to be directories.
         - dirsnotfound is a list of files that the dirstate thinks are
           directories and that were not found."""
 
         def badtype(mode):
             kind = _(b'unknown')
             if stat.S_ISCHR(mode):
                 kind = _(b'character device')
             elif stat.S_ISBLK(mode):
                 kind = _(b'block device')
             elif stat.S_ISFIFO(mode):
                 kind = _(b'fifo')
             elif stat.S_ISSOCK(mode):
                 kind = _(b'socket')
             elif stat.S_ISDIR(mode):
                 kind = _(b'directory')
             return _(b'unsupported file type (type is %s)') % kind
 
         badfn = match.bad
         dmap = self._map
         lstat = os.lstat
         getkind = stat.S_IFMT
         dirkind = stat.S_IFDIR
         regkind = stat.S_IFREG
         lnkkind = stat.S_IFLNK
         join = self._join
         dirsfound = []
         foundadd = dirsfound.append
         dirsnotfound = []
         notfoundadd = dirsnotfound.append
 
         if not match.isexact() and self._checkcase:
             normalize = self._normalize
         else:
             normalize = None
 
         files = sorted(match.files())
         subrepos.sort()
         i, j = 0, 0
         while i < len(files) and j < len(subrepos):
             subpath = subrepos[j] + b"/"
             if files[i] < subpath:
                 i += 1
                 continue
             while i < len(files) and files[i].startswith(subpath):
                 del files[i]
             j += 1
 
         if not files or b'' in files:
             files = [b'']
             # constructing the foldmap is expensive, so don't do it for the
             # common case where files is ['']
             normalize = None
         results = dict.fromkeys(subrepos)
         results[b'.hg'] = None
 
         for ff in files:
             if normalize:
                 nf = normalize(ff, False, True)
             else:
                 nf = ff
             if nf in results:
                 continue
 
             try:
                 st = lstat(join(nf))
                 kind = getkind(st.st_mode)
                 if kind == dirkind:
                     if nf in dmap:
                         # file replaced by dir on disk but still in dirstate
                         results[nf] = None
                     foundadd((nf, ff))
                 elif kind == regkind or kind == lnkkind:
                     results[nf] = st
                 else:
                     badfn(ff, badtype(kind))
                     if nf in dmap:
                         results[nf] = None
             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?
                     if self._map.hasdir(nf):
                         notfoundadd(nf)
                     else:
                         badfn(ff, encoding.strtolocal(inst.strerror))
 
         # match.files() may contain explicitly-specified paths that shouldn't
         # be taken; drop them from the list of files found. dirsfound/notfound
         # aren't filtered here because they will be tested later.
         if match.anypats():
             for f in list(results):
                 if f == b'.hg' or f in subrepos:
                     # keep sentinel to disable further out-of-repo walks
                     continue
                 if not match(f):
                     del results[f]
 
         # Case insensitive filesystems cannot rely on lstat() failing to detect
         # a case-only rename.  Prune the stat object for any file that does not
         # match the case in the filesystem, if there are multiple files that
         # normalize to the same path.
         if match.isexact() and self._checkcase:
             normed = {}
 
             for f, st in results.items():
                 if st is None:
                     continue
 
                 nc = util.normcase(f)
                 paths = normed.get(nc)
 
                 if paths is None:
                     paths = set()
                     normed[nc] = paths
 
                 paths.add(f)
 
             for norm, paths in normed.items():
                 if len(paths) > 1:
                     for path in paths:
                         folded = self._discoverpath(
                             path, norm, True, None, self._map.dirfoldmap
                         )
                         if path != folded:
                             results[path] = None
 
         return results, dirsfound, dirsnotfound
 
     def walk(self, match, subrepos, unknown, ignored, full=True):
         """
         Walk recursively through the directory tree, finding all files
         matched by match.
 
         If full is False, maybe skip some known-clean files.
 
         Return a dict mapping filename to stat-like object (either
         mercurial.osutil.stat instance or return value of os.stat()).
 
         """
         # full is a flag that extensions that hook into walk can use -- this
         # implementation doesn't use it at all. This satisfies the contract
         # because we only guarantee a "maybe".
 
         if ignored:
             ignore = util.never
             dirignore = util.never
         elif unknown:
             ignore = self._ignore
             dirignore = self._dirignore
         else:
             # if not unknown and not ignored, drop dir recursion and step 2
             ignore = util.always
             dirignore = util.always
 
         if self._sparsematchfn is not None:
             em = matchmod.exact(match.files())
             sm = matchmod.unionmatcher([self._sparsematcher, em])
             match = matchmod.intersectmatchers(match, sm)
 
         matchfn = match.matchfn
         matchalways = match.always()
         matchtdir = match.traversedir
         dmap = self._map
         listdir = util.listdir
         lstat = os.lstat
         dirkind = stat.S_IFDIR
         regkind = stat.S_IFREG
         lnkkind = stat.S_IFLNK
         join = self._join
 
         exact = skipstep3 = False
         if match.isexact():  # match.exact
             exact = True
             dirignore = util.always  # skip step 2
         elif match.prefix():  # match.match, no patterns
             skipstep3 = True
 
         if not exact and self._checkcase:
             normalize = self._normalize
             normalizefile = self._normalizefile
             skipstep3 = False
         else:
             normalize = self._normalize
             normalizefile = None
 
         # step 1: find all explicit files
         results, work, dirsnotfound = self._walkexplicit(match, subrepos)
         if matchtdir:
             for d in work:
                 matchtdir(d[0])
             for d in dirsnotfound:
                 matchtdir(d)
 
         skipstep3 = skipstep3 and not (work or dirsnotfound)
         work = [d for d in work if not dirignore(d[0])]
 
         # step 2: visit subdirectories
         def traverse(work, alreadynormed):
             wadd = work.append
             while work:
                 tracing.counter('dirstate.walk work', len(work))
                 nd = work.pop()
                 visitentries = match.visitchildrenset(nd)
                 if not visitentries:
                     continue
                 if visitentries == b'this' or visitentries == b'all':
                     visitentries = None
                 skip = None
                 if nd != b'':
                     skip = b'.hg'
                 try:
                     with tracing.log('dirstate.walk.traverse listdir %s', nd):
                         entries = listdir(join(nd), stat=True, skip=skip)
                 except (PermissionError, FileNotFoundError) as inst:
                     match.bad(
                         self.pathto(nd), encoding.strtolocal(inst.strerror)
                     )
                     continue
                 for f, kind, st in entries:
                     # Some matchers may return files in the visitentries set,
                     # instead of 'this', if the matcher explicitly mentions them
                     # and is not an exactmatcher. This is acceptable; we do not
                     # make any hard assumptions about file-or-directory below
                     # based on the presence of `f` in visitentries. If
                     # visitchildrenset returned a set, we can always skip the
                     # entries *not* in the set it provided regardless of whether
                     # they're actually a file or a directory.
                     if visitentries and f not in visitentries:
                         continue
                     if normalizefile:
                         # even though f might be a directory, we're only
                         # interested in comparing it to files currently in the
                         # dmap -- therefore normalizefile is enough
                         nf = normalizefile(
                             nd and (nd + b"/" + f) or f, True, True
                         )
                     else:
                         nf = nd and (nd + b"/" + f) or f
                     if nf not in results:
                         if kind == dirkind:
                             if not ignore(nf):
                                 if matchtdir:
                                     matchtdir(nf)
                                 wadd(nf)
                             if nf in dmap and (matchalways or matchfn(nf)):
                                 results[nf] = None
                         elif kind == regkind or kind == lnkkind:
                             if nf in dmap:
                                 if matchalways or matchfn(nf):
                                     results[nf] = st
                             elif (matchalways or matchfn(nf)) and not ignore(
                                 nf
                             ):
                                 # unknown file -- normalize if necessary
                                 if not alreadynormed:
                                     nf = normalize(nf, False, True)
                                 results[nf] = st
                         elif nf in dmap and (matchalways or matchfn(nf)):
                             results[nf] = None
 
         for nd, d in work:
             # alreadynormed means that processwork doesn't have to do any
             # expensive directory normalization
             alreadynormed = not normalize or nd == d
             traverse([d], alreadynormed)
 
         for s in subrepos:
             del results[s]
         del results[b'.hg']
 
         # step 3: visit remaining files from dmap
         if not skipstep3 and not exact:
             # If a dmap file is not in results yet, it was either
             # a) not matching matchfn b) ignored, c) missing, or d) under a
             # symlink directory.
             if not results and matchalways:
                 visit = [f for f in dmap]
             else:
                 visit = [f for f in dmap if f not in results and matchfn(f)]
             visit.sort()
 
             if unknown:
                 # unknown == True means we walked all dirs under the roots
                 # that wasn't ignored, and everything that matched was stat'ed
                 # and is already in results.
                 # The rest must thus be ignored or under a symlink.
                 audit_path = pathutil.pathauditor(self._root, cached=True)
 
                 for nf in iter(visit):
                     # If a stat for the same file was already added with a
                     # different case, don't add one for this, since that would
                     # make it appear as if the file exists under both names
                     # on disk.
                     if (
                         normalizefile
                         and normalizefile(nf, True, True) in results
                     ):
                         results[nf] = None
                     # Report ignored items in the dmap as long as they are not
                     # under a symlink directory.
                     elif audit_path.check(nf):
                         try:
                             results[nf] = lstat(join(nf))
                             # file was just ignored, no links, and exists
                         except OSError:
                             # file doesn't exist
                             results[nf] = None
                     else:
                         # It's either missing or under a symlink directory
                         # which we in this case report as missing
                         results[nf] = None
             else:
                 # We may not have walked the full directory tree above,
                 # so stat and check everything we missed.
                 iv = iter(visit)
                 for st in util.statfiles([join(i) for i in visit]):
                     results[next(iv)] = st
         return results
 
     def _rust_status(self, matcher, list_clean, list_ignored, list_unknown):
         if self._sparsematchfn is not None:
             em = matchmod.exact(matcher.files())
             sm = matchmod.unionmatcher([self._sparsematcher, em])
             matcher = matchmod.intersectmatchers(matcher, sm)
         # Force Rayon (Rust parallelism library) to respect the number of
         # workers. This is a temporary workaround until Rust code knows
         # how to read the config file.
         numcpus = self._ui.configint(b"worker", b"numcpus")
         if numcpus is not None:
             encoding.environ.setdefault(b'RAYON_NUM_THREADS', b'%d' % numcpus)
 
         workers_enabled = self._ui.configbool(b"worker", b"enabled", True)
         if not workers_enabled:
             encoding.environ[b"RAYON_NUM_THREADS"] = b"1"
 
         (
             lookup,
             modified,
             added,
             removed,
             deleted,
             clean,
             ignored,
             unknown,
             warnings,
             bad,
             traversed,
             dirty,
         ) = rustmod.status(
             self._map._map,
             matcher,
             self._rootdir,
             self._ignorefiles(),
             self._checkexec,
             bool(list_clean),
             bool(list_ignored),
             bool(list_unknown),
             bool(matcher.traversedir),
         )
 
         self._dirty |= dirty
 
         if matcher.traversedir:
             for dir in traversed:
                 matcher.traversedir(dir)
 
         if self._ui.warn:
             for item in warnings:
                 if isinstance(item, tuple):
                     file_path, syntax = item
                     msg = _(b"%s: ignoring invalid syntax '%s'\n") % (
                         file_path,
                         syntax,
                     )
                     self._ui.warn(msg)
                 else:
                     msg = _(b"skipping unreadable pattern file '%s': %s\n")
                     self._ui.warn(
                         msg
                         % (
                             pathutil.canonpath(
                                 self._rootdir, self._rootdir, item
                             ),
                             b"No such file or directory",
                         )
                     )
 
         for fn, message in bad:
             matcher.bad(fn, encoding.strtolocal(message))
 
         status = scmutil.status(
             modified=modified,
             added=added,
             removed=removed,
             deleted=deleted,
             unknown=unknown,
             ignored=ignored,
             clean=clean,
         )
         return (lookup, status)
 
     # XXX since this can make the dirstate dirty (through rust), we should
     # enforce that it is done withing an appropriate change-context that scope
     # the change and ensure it eventually get written on disk (or rolled back)
     def status(self, match, subrepos, ignored, clean, unknown):
         """Determine the status of the working copy relative to the
         dirstate and return a pair of (unsure, status), where status is of type
         scmutil.status and:
 
           unsure:
             files that might have been modified since the dirstate was
             written, but need to be read to be sure (size is the same
             but mtime differs)
           status.modified:
             files that have definitely been modified since the dirstate
             was written (different size or mode)
           status.clean:
             files that have definitely not been modified since the
             dirstate was written
         """
         listignored, listclean, listunknown = ignored, clean, unknown
         lookup, modified, added, unknown, ignored = [], [], [], [], []
         removed, deleted, clean = [], [], []
 
         dmap = self._map
         dmap.preload()
 
         use_rust = True
 
         allowed_matchers = (
             matchmod.alwaysmatcher,
             matchmod.differencematcher,
             matchmod.exactmatcher,
             matchmod.includematcher,
             matchmod.intersectionmatcher,
             matchmod.nevermatcher,
             matchmod.unionmatcher,
         )
 
         if rustmod is None:
             use_rust = False
         elif self._checkcase:
             # Case-insensitive filesystems are not handled yet
             use_rust = False
         elif subrepos:
             use_rust = False
         elif not isinstance(match, allowed_matchers):
             # Some matchers have yet to be implemented
             use_rust = False
 
         # Get the time from the filesystem so we can disambiguate files that
         # appear modified in the present or future.
         try:
             mtime_boundary = timestamp.get_fs_now(self._opener)
         except OSError:
             # In largefiles or readonly context
             mtime_boundary = None
 
         if use_rust:
             try:
                 res = self._rust_status(
                     match, listclean, listignored, listunknown
                 )
                 return res + (mtime_boundary,)
             except rustmod.FallbackError:
                 pass
 
         def noop(f):
             pass
 
         dcontains = dmap.__contains__
         dget = dmap.__getitem__
         ladd = lookup.append  # aka "unsure"
         madd = modified.append
         aadd = added.append
         uadd = unknown.append if listunknown else noop
         iadd = ignored.append if listignored else noop
         radd = removed.append
         dadd = deleted.append
         cadd = clean.append if listclean else noop
         mexact = match.exact
         dirignore = self._dirignore
         checkexec = self._checkexec
         checklink = self._checklink
         copymap = self._map.copymap
 
         # We need to do full walks when either
         # - we're listing all clean files, or
         # - match.traversedir does something, because match.traversedir should
         #   be called for every dir in the working dir
         full = listclean or match.traversedir is not None
         for fn, st in self.walk(
             match, subrepos, listunknown, listignored, full=full
         ).items():
             if not dcontains(fn):
                 if (listignored or mexact(fn)) and dirignore(fn):
                     if listignored:
                         iadd(fn)
                 else:
                     uadd(fn)
                 continue
 
             t = dget(fn)
             mode = t.mode
             size = t.size
 
             if not st and t.tracked:
                 dadd(fn)
             elif t.p2_info:
                 madd(fn)
             elif t.added:
                 aadd(fn)
             elif t.removed:
                 radd(fn)
             elif t.tracked:
                 if not checklink and t.has_fallback_symlink:
                     # If the file system does not support symlink, the mode
                     # might not be correctly stored in the dirstate, so do not
                     # trust it.
                     ladd(fn)
                 elif not checkexec and t.has_fallback_exec:
                     # If the file system does not support exec bits, the mode
                     # might not be correctly stored in the dirstate, so do not
                     # trust it.
                     ladd(fn)
                 elif (
                     size >= 0
                     and (
                         (size != st.st_size and size != st.st_size & _rangemask)
                         or ((mode ^ st.st_mode) & 0o100 and checkexec)
                     )
                     or fn in copymap
                 ):
                     if stat.S_ISLNK(st.st_mode) and size != st.st_size:
                         # issue6456: Size returned may be longer due to
                         # encryption on EXT-4 fscrypt, undecided.
                         ladd(fn)
                     else:
                         madd(fn)
                 elif not t.mtime_likely_equal_to(timestamp.mtime_of(st)):
                     # There might be a change in the future if for example the
                     # internal clock is off, but this is a case where the issues
                     # the user would face would be a lot worse and there is
                     # nothing we can really do.
                     ladd(fn)
                 elif listclean:
                     cadd(fn)
         status = scmutil.status(
             modified, added, removed, deleted, unknown, ignored, clean
         )
         return (lookup, status, mtime_boundary)
 
     def matches(self, match):
         """
         return files in the dirstate (in whatever state) filtered by match
         """
         dmap = self._map
         if rustmod is not None:
             dmap = self._map._map
 
         if match.always():
             return dmap.keys()
         files = match.files()
         if match.isexact():
             # fast path -- filter the other way around, since typically files is
             # much smaller than dmap
             return [f for f in files if f in dmap]
         if match.prefix() and all(fn in dmap for fn in files):
             # fast path -- all the values are known to be files, so just return
             # that
             return list(files)
         return [f for f in dmap if match(f)]
 
     def _actualfilename(self, tr):
         if tr:
             return self._pendingfilename
         else:
             return self._filename
 
     def all_file_names(self):
         """list all filename currently used by this dirstate
 
         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,)
 
     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"
         )
         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():
             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:
             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))

tests/test-largefiles-update.t

@@ -1,828 +1,842 b''
 #require no-reposimplestore
 
 This file focuses mainly on updating largefiles in the working
 directory (and ".hg/largefiles/dirstate")
 
   $ cat >> $HGRCPATH <<EOF
   > [ui]
   > merge = internal:merge
   > [extensions]
   > largefiles =
   > [extdiff]
   > # for portability:
   > pdiff = sh "$RUNTESTDIR/pdiff"
   > EOF
 
   $ hg init repo
   $ cd repo
 
   $ echo large1 > large1
   $ echo large2 > large2
   $ hg add --large large1 large2
   $ echo normal1 > normal1
   $ hg add normal1
   $ hg commit -m '#0'
   $ echo 'large1 in #1' > large1
   $ echo 'normal1 in #1' > normal1
   $ hg commit -m '#1'
   $ hg pdiff -r '.^' --config extensions.extdiff=
   diff -Nru repo.0d9d9b8dc9a3/.hglf/large1 repo/.hglf/large1
   --- repo.0d9d9b8dc9a3/.hglf/large1	* (glob)
   +++ repo/.hglf/large1	* (glob)
   @@ -1* +1* @@ (glob)
   -4669e532d5b2c093a78eca010077e708a071bb64
   +58e24f733a964da346e2407a2bee99d9001184f5
   diff -Nru repo.0d9d9b8dc9a3/normal1 repo/normal1
   --- repo.0d9d9b8dc9a3/normal1	* (glob)
   +++ repo/normal1	* (glob)
   @@ -1* +1* @@ (glob)
   -normal1
   +normal1 in #1
   [1]
   $ hg update -q -C 0
   $ echo 'large2 in #2' > large2
   $ hg commit -m '#2'
   created new head
 
 Test that update also updates the lfdirstate of 'unsure' largefiles after
 hashing them:
 
 The previous operations will usually have left us with largefiles with a mtime
 within the same second as the dirstate was written.
 The lfdirstate entries will thus have been written with an invalidated/unset
 mtime to make sure further changes within the same second is detected.
 We will however occasionally be "lucky" and get a tick between writing
 largefiles and writing dirstate so we get valid lfdirstate timestamps. The
 following verification is thus disabled but can be verified manually.
 
 #if false
   $ hg debugdirstate --large --nodate
   n 644          7 unset               large1
   n 644         13 unset               large2
 #endif
 
 Wait to make sure we get a tick so the mtime of the largefiles become valid.
 
   $ sleep 1
 
 A linear merge will update standins before performing the actual merge. It will
 do a lfdirstate status walk and find 'unset'/'unsure' files, hash them, and
 update the corresponding standins.
 
 Verify that it actually marks the clean files as clean in lfdirstate so
 we don't have to hash them again next time we update.
 
 # note:
 #    We do this less agressively now, to avoid race condition, however the
 #    cache
 #    is properly set after the next status
 #
 #    The "changed" output is marked as missing-correct-output/known-bad-output
 #    for clarify
 
   $ hg up
   0 files updated, 0 files merged, 0 files removed, 0 files unresolved
   updated to "f74e50bd9e55: #2"
   1 other heads for branch "default"
   $ hg debugdirstate --large --nodate
   n 644          7 set                 large1 (missing-correct-output !)
   n 644         13 set                 large2 (missing-correct-output !)
   n   0         -1 unset               large1 (known-bad-output !)
   n   0         -1 unset               large2 (known-bad-output !)
   $ sleep 1 # so that mtime are not ambiguous
   $ hg status
   $ hg debugdirstate --large --nodate
   n 644          7 set                 large1
   n 644         13 set                 large2
 
 Test that lfdirstate keeps track of last modification of largefiles and
 prevents unnecessary hashing of content - also after linear/noop update
 
 (XXX Since there is a possible race during update, we only do this after the next
 status call, this is slower, but more correct)
 
   $ sleep 1
   $ hg st
   $ hg debugdirstate --large --nodate
   n 644          7 set                 large1
   n 644         13 set                 large2
   $ hg up
   0 files updated, 0 files merged, 0 files removed, 0 files unresolved
   updated to "f74e50bd9e55: #2"
   1 other heads for branch "default"
   $ hg debugdirstate --large --nodate
   n 644          7 set                 large1 (missing-correct-output !)
   n 644         13 set                 large2 (missing-correct-output !)
   n   0         -1 unset               large1 (known-bad-output !)
   n   0         -1 unset               large2 (known-bad-output !)
   $ sleep 1 # so that mtime are not ambiguous
   $ hg status
   $ hg debugdirstate --large --nodate
   n 644          7 set                 large1
   n 644         13 set                 large2
 
 Test that "hg merge" updates largefiles from "other" correctly
 
 (getting largefiles from "other" normally)
 
   $ hg status -A large1
   C large1
   $ cat large1
   large1
   $ cat .hglf/large1
   4669e532d5b2c093a78eca010077e708a071bb64
   $ hg merge --config debug.dirstate.delaywrite=2
   getting changed largefiles
   1 largefiles updated, 0 removed
   2 files updated, 0 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg status -A large1
   M large1
   $ cat large1
   large1 in #1
   $ cat .hglf/large1
   58e24f733a964da346e2407a2bee99d9001184f5
   $ hg diff -c 1 --nodates .hglf/large1 | grep '^[+-][0-9a-z]'
   -4669e532d5b2c093a78eca010077e708a071bb64
   +58e24f733a964da346e2407a2bee99d9001184f5
 
 (getting largefiles from "other" via conflict prompt)
 
   $ hg update -q -C 2
   $ echo 'large1 in #3' > large1
   $ echo 'normal1 in #3' > normal1
   $ hg commit -m '#3'
   $ cat .hglf/large1
   e5bb990443d6a92aaf7223813720f7566c9dd05b
   $ hg merge --config debug.dirstate.delaywrite=2 --config ui.interactive=True <<EOF
   > o
   > EOF
   largefile large1 has a merge conflict
   ancestor was 4669e532d5b2c093a78eca010077e708a071bb64
   you can keep (l)ocal e5bb990443d6a92aaf7223813720f7566c9dd05b or take (o)ther 58e24f733a964da346e2407a2bee99d9001184f5.
   what do you want to do? o
   merging normal1
   warning: conflicts while merging normal1! (edit, then use 'hg resolve --mark')
   getting changed largefiles
   1 largefiles updated, 0 removed
   0 files updated, 1 files merged, 0 files removed, 1 files unresolved
   use 'hg resolve' to retry unresolved file merges or 'hg merge --abort' to abandon
   [1]
   $ hg status -A large1
   M large1
   $ cat large1
   large1 in #1
   $ cat .hglf/large1
   58e24f733a964da346e2407a2bee99d9001184f5
   $ rm normal1.orig
 
 (merge non-existing largefiles from "other" via conflict prompt -
 make sure the following commit doesn't abort in a confusing way when trying to
 mark the non-existing file as normal in lfdirstate)
 
   $ mv .hg/largefiles/58e24f733a964da346e2407a2bee99d9001184f5 .
   $ hg update -q -C 3
   $ hg merge --config largefiles.usercache=not --config debug.dirstate.delaywrite=2 --tool :local --config ui.interactive=True <<EOF
   > o
   > EOF
   largefile large1 has a merge conflict
   ancestor was 4669e532d5b2c093a78eca010077e708a071bb64
   you can keep (l)ocal e5bb990443d6a92aaf7223813720f7566c9dd05b or take (o)ther 58e24f733a964da346e2407a2bee99d9001184f5.
   what do you want to do? o
   getting changed largefiles
   large1: largefile 58e24f733a964da346e2407a2bee99d9001184f5 not available from file:/*/$TESTTMP/repo (glob)
   0 largefiles updated, 0 removed
   0 files updated, 2 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg commit -m '1-2-3 testing' --config largefiles.usercache=not
   large1: largefile 58e24f733a964da346e2407a2bee99d9001184f5 not available from local store
   $ hg up -C . --config largefiles.usercache=not
   getting changed largefiles
   large1: largefile 58e24f733a964da346e2407a2bee99d9001184f5 not available from file:/*/$TESTTMP/repo (glob)
   0 largefiles updated, 0 removed
   0 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg st large1
   ! large1
   $ hg rollback -q
   $ mv 58e24f733a964da346e2407a2bee99d9001184f5 .hg/largefiles/
 
 Test that "hg revert -r REV" updates largefiles from "REV" correctly
 
   $ hg update -q -C 3
   $ hg status -A large1
   C large1
   $ cat large1
   large1 in #3
   $ cat .hglf/large1
   e5bb990443d6a92aaf7223813720f7566c9dd05b
   $ hg diff -c 1 --nodates .hglf/large1 | grep '^[+-][0-9a-z]'
   -4669e532d5b2c093a78eca010077e708a071bb64
   +58e24f733a964da346e2407a2bee99d9001184f5
   $ hg revert --no-backup -r 1 --config debug.dirstate.delaywrite=2 large1
   $ hg status -A large1
   M large1
   $ cat large1
   large1 in #1
   $ cat .hglf/large1
   58e24f733a964da346e2407a2bee99d9001184f5
 
 Test that "hg rollback" restores status of largefiles correctly
 
   $ hg update -C -q
   $ hg remove large1
   $ test -f .hglf/large1
   [1]
   $ hg forget large2
   $ test -f .hglf/large2
   [1]
   $ echo largeX > largeX
   $ hg add --large largeX
   $ cat .hglf/largeX
   
   $ hg commit -m 'will be rollback-ed soon'
   $ echo largeY > largeY
   $ hg add --large largeY
 
   $ hg status -A large1
   large1: $ENOENT$
 
   $ hg status -A large2
   ? large2
   $ hg status -A largeX
   C largeX
   $ hg status -A largeY
   A largeY
   $ hg rollback
   repository tip rolled back to revision 3 (undo commit)
   working directory now based on revision 3
   $ hg status -A large1
   R large1
   $ test -f .hglf/large1
   [1]
   $ hg status -A large2
   R large2
   $ test -f .hglf/large2
   [1]
   $ hg status -A largeX
   A largeX
   $ cat .hglf/largeX
   
   $ hg status -A largeY
   ? largeY
   $ test -f .hglf/largeY
   [1]
   $ rm largeY
 
 Test that "hg rollback" restores standins correctly
 
   $ hg commit -m 'will be rollback-ed soon'
   $ hg update -q -C 2
   $ cat large1
   large1
   $ cat .hglf/large1
   4669e532d5b2c093a78eca010077e708a071bb64
   $ cat large2
   large2 in #2
   $ cat .hglf/large2
   3cfce6277e7668985707b6887ce56f9f62f6ccd9
 
   $ hg rollback -q -f
   $ cat large1
   large1
   $ cat .hglf/large1
   4669e532d5b2c093a78eca010077e708a071bb64
   $ cat large2
   large2 in #2
   $ cat .hglf/large2
   3cfce6277e7668985707b6887ce56f9f62f6ccd9
 
 (rollback the parent of the working directory, when the parent of it
 is not branch-tip)
 
   $ hg update -q -C 1
   $ cat .hglf/large1
   58e24f733a964da346e2407a2bee99d9001184f5
   $ cat .hglf/large2
   1deebade43c8c498a3c8daddac0244dc55d1331d
 
   $ echo normalX > normalX
   $ hg add normalX
   $ hg commit -m 'will be rollback-ed soon'
   $ hg rollback -q
 
   $ cat .hglf/large1
   58e24f733a964da346e2407a2bee99d9001184f5
   $ cat .hglf/large2
   1deebade43c8c498a3c8daddac0244dc55d1331d
   $ rm normalX
 
 Test that "hg status" shows status of largefiles correctly just after
 automated commit like rebase/transplant
 
   $ cat >> .hg/hgrc <<EOF
   > [extensions]
   > rebase =
   > strip =
   > transplant =
   > EOF
   $ hg update -q -C 1
   $ hg remove large1
   $ echo largeX > largeX
   $ hg add --large largeX
   $ hg commit -m '#4'
 
   $ hg rebase -s 1 -d 2 --keep
   rebasing 1:72518492caa6 "#1"
   rebasing 4:07d6153b5c04 tip "#4"
 
   $ hg status -A large1
   large1: $ENOENT$
 
   $ hg status -A largeX
   C largeX
   $ hg strip -q 5
 
   $ hg update -q -C 2
   $ hg transplant -q 1 4
 
   $ hg status -A large1
   large1: $ENOENT$
 
   $ hg status -A largeX
   C largeX
   $ hg strip -q 5
 
   $ hg update -q -C 2
   $ hg transplant -q --merge 1 --merge 4
 
   $ hg status -A large1
   large1: $ENOENT$
 
   $ hg status -A largeX
   C largeX
   $ hg strip -q 5
 
 Test that linear merge can detect modification (and conflict) correctly
 
 (linear merge without conflict)
 
   $ echo 'large2 for linear merge (no conflict)' > large2
   $ hg update 3 --config debug.dirstate.delaywrite=2
   getting changed largefiles
   1 largefiles updated, 0 removed
   2 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg status -A large2
   M large2
   $ cat large2
   large2 for linear merge (no conflict)
   $ cat .hglf/large2
   9c4bf8f1b33536d6e5f89447e10620cfe52ea710
 
 (linear merge with conflict, choosing "other")
 
   $ hg update -q -C 2
   $ echo 'large1 for linear merge (conflict)' > large1
   $ hg update 3 --config ui.interactive=True <<EOF
   > o
   > EOF
   largefile large1 has a merge conflict
   ancestor was 4669e532d5b2c093a78eca010077e708a071bb64
   you can keep (l)ocal ba94c2efe5b7c5e0af8d189295ce00553b0612b7 or take (o)ther e5bb990443d6a92aaf7223813720f7566c9dd05b.
   what do you want to do? o
   getting changed largefiles
   1 largefiles updated, 0 removed
   1 files updated, 1 files merged, 0 files removed, 0 files unresolved
   $ hg status -A large1
   C large1
   $ cat large1
   large1 in #3
   $ cat .hglf/large1
   e5bb990443d6a92aaf7223813720f7566c9dd05b
 
 (linear merge with conflict, choosing "local")
 
   $ hg update -q -C 2
   $ echo 'large1 for linear merge (conflict)' > large1
   $ hg update 3 --config debug.dirstate.delaywrite=2
   largefile large1 has a merge conflict
   ancestor was 4669e532d5b2c093a78eca010077e708a071bb64
   you can keep (l)ocal ba94c2efe5b7c5e0af8d189295ce00553b0612b7 or take (o)ther e5bb990443d6a92aaf7223813720f7566c9dd05b.
   what do you want to do? l
   1 files updated, 1 files merged, 0 files removed, 0 files unresolved
   $ hg status -A large1
   M large1
   $ cat large1
   large1 for linear merge (conflict)
   $ cat .hglf/large1
   ba94c2efe5b7c5e0af8d189295ce00553b0612b7
 
 Test a linear merge to a revision containing same-name normal file
 
   $ hg update -q -C 3
   $ hg remove large2
   $ echo 'large2 as normal file' > large2
   $ hg add large2
   $ echo 'large3 as normal file' > large3
   $ hg add large3
   $ hg commit -m '#5'
   $ hg manifest
   .hglf/large1
   large2
   large3
   normal1
 
 (modified largefile is already switched to normal)
 
   $ hg update -q -C 2
   $ echo 'modified large2 for linear merge' > large2
   $ hg update -q 5
   remote turned local largefile large2 into a normal file
   keep (l)argefile or use (n)ormal file? l
   $ hg debugdirstate --no-dates | grep large2
   a   0         -1 unset               .hglf/large2
   r   0          0 set                 large2
   $ hg status -A large2
   A large2
   $ cat large2
   modified large2 for linear merge
 
 (added largefile is already committed as normal)
 
   $ hg update -q -C 2
   $ echo 'large3 as large file for linear merge' > large3
   $ hg add --large large3
   $ hg update -q 5
   remote turned local largefile large3 into a normal file
   keep (l)argefile or use (n)ormal file? l
   $ hg debugdirstate --no-dates | grep large3
   a   0         -1 unset               .hglf/large3
   r   0          0 set                 large3
   $ hg status -A large3
   A large3
   $ cat large3
   large3 as large file for linear merge
   $ rm -f large3 .hglf/large3
 
 Test that the internal linear merging works correctly
 (both heads are stripped to keep pairing of revision number and commit log)
 
   $ hg update -q -C 2
   $ hg strip 3 4
   saved backup bundle to $TESTTMP/repo/.hg/strip-backup/9530e27857f7-2e7b195d-backup.hg
   $ mv .hg/strip-backup/9530e27857f7-2e7b195d-backup.hg $TESTTMP
 
 (internal linear merging at "hg pull --update")
 
   $ echo 'large1 for linear merge (conflict)' > large1
   $ echo 'large2 for linear merge (conflict with normal file)' > large2
   $ hg pull --update --config debug.dirstate.delaywrite=2 $TESTTMP/9530e27857f7-2e7b195d-backup.hg
   pulling from $TESTTMP/9530e27857f7-2e7b195d-backup.hg
   searching for changes
   adding changesets
   adding manifests
   adding file changes
   added 3 changesets with 5 changes to 5 files
   new changesets 9530e27857f7:d65e59e952a9 (3 drafts)
   remote turned local largefile large2 into a normal file
   keep (l)argefile or use (n)ormal file? l
   largefile large1 has a merge conflict
   ancestor was 4669e532d5b2c093a78eca010077e708a071bb64
   you can keep (l)ocal ba94c2efe5b7c5e0af8d189295ce00553b0612b7 or take (o)ther e5bb990443d6a92aaf7223813720f7566c9dd05b.
   what do you want to do? l
   2 files updated, 1 files merged, 0 files removed, 0 files unresolved
   updated to "d65e59e952a9: #5"
   1 other heads for branch "default"
 
   $ hg status -A large1
   M large1
   $ cat large1
   large1 for linear merge (conflict)
   $ cat .hglf/large1
   ba94c2efe5b7c5e0af8d189295ce00553b0612b7
   $ hg status -A large2
   A large2
   $ cat large2
   large2 for linear merge (conflict with normal file)
   $ cat .hglf/large2
   d7591fe9be0f6227d90bddf3e4f52ff41fc1f544
 
 (internal linear merging at "hg unbundle --update")
 
   $ hg update -q -C 2
   $ hg rollback -q
 
   $ echo 'large1 for linear merge (conflict)' > large1
   $ echo 'large2 for linear merge (conflict with normal file)' > large2
   $ hg unbundle --update --config debug.dirstate.delaywrite=2 $TESTTMP/9530e27857f7-2e7b195d-backup.hg
   adding changesets
   adding manifests
   adding file changes
   added 3 changesets with 5 changes to 5 files
   new changesets 9530e27857f7:d65e59e952a9 (3 drafts)
   remote turned local largefile large2 into a normal file
   keep (l)argefile or use (n)ormal file? l
   largefile large1 has a merge conflict
   ancestor was 4669e532d5b2c093a78eca010077e708a071bb64
   you can keep (l)ocal ba94c2efe5b7c5e0af8d189295ce00553b0612b7 or take (o)ther e5bb990443d6a92aaf7223813720f7566c9dd05b.
   what do you want to do? l
   2 files updated, 1 files merged, 0 files removed, 0 files unresolved
   updated to "d65e59e952a9: #5"
   1 other heads for branch "default"
 
   $ hg status -A large1
   M large1
   $ cat large1
   large1 for linear merge (conflict)
   $ cat .hglf/large1
   ba94c2efe5b7c5e0af8d189295ce00553b0612b7
   $ hg status -A large2
   A large2
   $ cat large2
   large2 for linear merge (conflict with normal file)
   $ cat .hglf/large2
   d7591fe9be0f6227d90bddf3e4f52ff41fc1f544
 
 (internal linear merging in subrepo at "hg update")
 
   $ cd ..
   $ hg init subparent
   $ cd subparent
 
   $ hg clone -q -u 2 ../repo sub
   $ cat > .hgsub <<EOF
   > sub = sub
   > EOF
   $ hg add .hgsub
   $ hg commit -m '#0@parent'
   $ cat .hgsubstate
   f74e50bd9e5594b7cf1e6c5cbab86ddd25f3ca2f sub
   $ hg -R sub update -q
   $ hg commit -m '#1@parent'
   $ cat .hgsubstate
   d65e59e952a9638e2ce863b41a420ca723dd3e8d sub
   $ hg update -q 0
 
   $ echo 'large1 for linear merge (conflict)' > sub/large1
   $ echo 'large2 for linear merge (conflict with normal file)' > sub/large2
   $ hg update --config ui.interactive=True --config debug.dirstate.delaywrite=2 <<EOF
   > m
   > r
   > l
   > l
   > EOF
    subrepository sub diverged (local revision: f74e50bd9e55, remote revision: d65e59e952a9)
   you can (m)erge, keep (l)ocal [working copy] or keep (r)emote [destination].
   what do you want to do? m
    subrepository sources for sub differ (in checked out version)
   you can use (l)ocal source (f74e50bd9e55) or (r)emote source (d65e59e952a9).
   what do you want to do? r
   remote turned local largefile large2 into a normal file
   keep (l)argefile or use (n)ormal file? l
   largefile large1 has a merge conflict
   ancestor was 4669e532d5b2c093a78eca010077e708a071bb64
   you can keep (l)ocal ba94c2efe5b7c5e0af8d189295ce00553b0612b7 or take (o)ther e5bb990443d6a92aaf7223813720f7566c9dd05b.
   what do you want to do? l
   2 files updated, 1 files merged, 0 files removed, 0 files unresolved
   0 files updated, 0 files merged, 0 files removed, 0 files unresolved
 
   $ hg -R sub status -A sub/large1
   M sub/large1
   $ cat sub/large1
   large1 for linear merge (conflict)
   $ cat sub/.hglf/large1
   ba94c2efe5b7c5e0af8d189295ce00553b0612b7
   $ hg -R sub status -A sub/large2
   A sub/large2
   $ cat sub/large2
   large2 for linear merge (conflict with normal file)
   $ cat sub/.hglf/large2
   d7591fe9be0f6227d90bddf3e4f52ff41fc1f544
 
   $ cd ..
   $ cd repo
 
 Test that rebase updates largefiles in the working directory even if
 it is aborted by conflict.
 
   $ hg update -q -C 3
   $ cat .hglf/large1
   e5bb990443d6a92aaf7223813720f7566c9dd05b
   $ cat large1
   large1 in #3
   $ hg rebase -s 1 -d 3 --keep --config ui.interactive=True <<EOF
   > o
   > EOF
   rebasing 1:72518492caa6 "#1"
   largefile large1 has a merge conflict
   ancestor was 4669e532d5b2c093a78eca010077e708a071bb64
   you can keep (l)ocal e5bb990443d6a92aaf7223813720f7566c9dd05b or take (o)ther 58e24f733a964da346e2407a2bee99d9001184f5.
   what do you want to do? o
   merging normal1
   warning: conflicts while merging normal1! (edit, then use 'hg resolve --mark')
   unresolved conflicts (see 'hg resolve', then 'hg rebase --continue')
   [240]
   $ cat .hglf/large1
   58e24f733a964da346e2407a2bee99d9001184f5
   $ cat large1
   large1 in #1
   $ rm normal1.orig
 
 Test that rebase updates standins for manually modified largefiles at
 the 1st commit of resuming.
 
   $ echo "manually modified before 'hg rebase --continue'" > large1
   $ hg resolve -m normal1
   (no more unresolved files)
   continue: hg rebase --continue
   $ hg rebase --continue --config ui.interactive=True <<EOF
   > c
   > EOF
   rebasing 1:72518492caa6 "#1"
   rebasing 4:07d6153b5c04 "#4"
   file '.hglf/large1' was deleted in other [source] but was modified in local [dest].
   You can use (c)hanged version, (d)elete, or leave (u)nresolved.
   What do you want to do? c
 
   $ hg diff -c "tip~1" --nodates .hglf/large1 | grep '^[+-][0-9a-z]'
   -e5bb990443d6a92aaf7223813720f7566c9dd05b
   +8a4f783556e7dea21139ca0466eafce954c75c13
   $ rm -f large1
   $ hg update -q -C tip
   $ cat large1
   manually modified before 'hg rebase --continue'
 
 Test that transplant updates largefiles, of which standins are safely
 changed, even if it is aborted by conflict of other.
 
   $ hg update -q -C 5
   $ cat .hglf/large1
   e5bb990443d6a92aaf7223813720f7566c9dd05b
   $ cat large1
   large1 in #3
   $ hg diff -c 4 .hglf/largeX | grep '^[+-][0-9a-z]'
   +fa44618ea25181aff4f48b70428294790cec9f61
   $ hg transplant 4
   applying 07d6153b5c04
   patching file .hglf/large1
   Hunk #1 FAILED at 0
   1 out of 1 hunks FAILED -- saving rejects to file .hglf/large1.rej
   patch failed to apply
   abort: fix up the working directory and run hg transplant --continue
   [255]
   $ hg status -A large1
   C large1
   $ cat .hglf/large1
   e5bb990443d6a92aaf7223813720f7566c9dd05b
   $ cat large1
   large1 in #3
   $ hg status -A largeX
   A largeX
   $ cat .hglf/largeX
   fa44618ea25181aff4f48b70428294790cec9f61
   $ cat largeX
   largeX
 
 Test that transplant updates standins for manually modified largefiles
 at the 1st commit of resuming.
 
   $ echo "manually modified before 'hg transplant --continue'" > large1
   $ hg transplant --continue
   07d6153b5c04 transplanted as f1bf30eb88cc
   $ hg diff -c tip .hglf/large1 | grep '^[+-][0-9a-z]'
   -e5bb990443d6a92aaf7223813720f7566c9dd05b
   +6a4f36d4075fbe0f30ec1d26ca44e63c05903671
   $ rm -f large1
   $ hg update -q -C tip
   $ cat large1
   manually modified before 'hg transplant --continue'
 
 Test that "hg status" doesn't show removal of largefiles not managed
 in the target context.
 
   $ hg update -q -C 4
   $ hg remove largeX
   $ hg status -A largeX
   R largeX
   $ hg status -A --rev '.^1' largeX
 
 #if execbit
 
 Test that "hg status" against revisions other than parent notices exec
 bit changes of largefiles.
 
   $ hg update -q -C 4
 
 (the case that large2 doesn't have exec bit in the target context but
 in the working context)
 
   $ chmod +x large2
   $ hg status -A --rev 0 large2
   M large2
   $ hg commit -m 'chmod +x large2'
 
 (the case that large2 has exec bit in the target context but not in
 the working context)
 
   $ echo dummy > dummy
   $ hg add dummy
   $ hg commit -m 'revision for separation'
   $ chmod -x large2
   $ hg status -A --rev '.^1' large2
   M large2
 
 #else
 
 Test that "hg status" against revisions other than parent ignores exec
 bit correctly on the platform being unaware of it.
 
   $ hg update -q -C 4
 
   $ cat > ../exec-bit.patch <<EOF
   > # HG changeset patch
   > # User test
   > # Date 0 0
   > #      Thu Jan 01 00:00:00 1970 +0000
   > # Node ID be1b433a65b12b27b5519d92213e14f7e1769b90
   > # Parent  07d6153b5c04313efb75deec9ba577de7faeb727
   > chmod +x large2
   > 
   > diff --git a/.hglf/large2 b/.hglf/large2
   > old mode 100644
   > new mode 100755
   > EOF
   $ hg import --exact --bypass ../exec-bit.patch
   applying ../exec-bit.patch
   $ hg status -A --rev tip large2
   C large2
 
 #endif
 
 The fileset revset is evaluated for each revision, instead of once on wdir(),
 and then patterns matched on each revision.  Here, no exec bits are set in
 wdir(), but a matching revision is detected.
 
 (Teach large2 is not an executable. Maybe this is a bug of largefiles.)
 #if execbit
   $ chmod -x .hglf/large2
 #endif
 
   $ hg files 'set:exec()'
   [1]
   $ hg log -qr 'file("set:exec()")'
   9:be1b433a65b1
 
-Test a fatal error interrupting an update. Verify that status report dirty
-files correctly after an interrupted update. Also verify that checking all
-hashes reveals it isn't clean.
+Test a fatal error interrupting an update
+-----------------------------------------
+
+In a previous version this test was tasked to:
+| verify that status report dirty files correctly after an interrupted
+| update. Also verify that checking all hashes reveals it isn't clean.
+
+In the mean time improvement to the update logic means it is much harder to get the dirstate file written too early. So the original intend seems "fine".
+
+However, it shows another error where the standin file for large1 seems to be
+silently updated, confusing the general logic. This seems to have been broken
+before our updates and the test is marked as such.
 
 Start with clean dirstates:
   $ hg up --quiet --clean --rev "8^"
   $ sleep 1
   $ cat large1
   large1 in #3
   $ hg st
 
 Update standins without updating largefiles - large1 is modified and largeX is
 added:
   $ cat << EOF > ../crashupdatelfiles.py
   > import hgext.largefiles.lfutil
   > def getlfilestoupdate(oldstandins, newstandins):
   >      raise SystemExit(7)
   > hgext.largefiles.lfutil.getlfilestoupdate = getlfilestoupdate
   > EOF
   $ hg up -Cr "8" --config extensions.crashupdatelfiles=../crashupdatelfiles.py
   [254]
 Check large1 content and status ... and that update will undo modifications:
   $ hg id
-  f1bf30eb88cc
+  d65e59e952a9+ (known-bad-output !)
+  d65e59e952a9 (missing-correct-output !)
   $ cat large1
   large1 in #3
   $ hg st
-  M large1
-  ! largeX
   $ hg up -Cr 8
   getting changed largefiles
-  2 largefiles updated, 0 removed
+  1 largefiles updated, 0 removed (known-bad-output !)
+  2 largefiles updated, 0 removed (missing-correct-output !)
   2 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ cat large1
-  manually modified before 'hg transplant --continue'
+  large1 in #3 (known-bad-output !)
+  manually modified before 'hg transplant --continue' (missing-correct-output !)
   $ hg st
+  M large1 (known-bad-output !)
+
+  $ hg revert --all --no-backup
+  reverting .hglf/large1 (known-bad-output !)
 Force largefiles rehashing and check that all changes have been caught by
 status and update:
   $ rm .hg/largefiles/dirstate
   $ hg st
 
   $ cd ..
 
 Test that "hg convert" avoids copying largefiles from the working
 directory into store, because "hg convert" doesn't update largefiles
 in the working directory (removing files under ".cache/largefiles"
 forces "hg convert" to copy corresponding largefiles)
 
   $ cat >> $HGRCPATH <<EOF
   > [extensions]
   > convert =
   > EOF
 
   $ rm $TESTTMP/.cache/largefiles/6a4f36d4075fbe0f30ec1d26ca44e63c05903671
   $ hg convert -q repo repo.converted