dirstate: add a way to test races happening during status

We add the `devel.sync.status.pre-dirstate-write-file` config option to easily
test what happens when other operations happen during the window where
`hg status` is done working but has not updated the cache on disk yet.

We introduce the framework for testing such races too, actual tests will be
added in the next changesets. For now the test is only checking dirstate-v1. We
will extend the test coverage later too.

Check test documentation for details.

Code change from Raphaël Gomès <[e-mail]>
Test change from Pierre-Yves David <[e-mail]>

tests/test-dirstate-status-write-race.t

@@ -0,0 +1,190 b''
+=====================================================================
+Check potential race conditions between a status and other operations
+=====================================================================
+
+The `hg status` command can run without the wlock, however it might end up
+having to update the on-disk dirstate files, for example to mark ambiguous
+files as clean, or to update directory caches information with dirstate-v2.
+
+
+If another process updates the dirstate in the meantime we might run into
+trouble. Especially, commands doing semantic changes like `hg add` or
+ `hg commit` should not see their update erased by a concurrent status.
+
+Unlike commands like `add` or `commit`, `status` only writes the dirstate
+to update caches, no actual information is lost if we fail to write to disk.
+
+
+This test file is meant to test various cases where such parallel operations
+between a status with reasons to update the dirstate and another semantic
+changes happen.
+
+
+Setup
+=====
+
+  $ directories="dir dir/nested dir2"
+  $ first_files="dir/nested/a dir/b dir/c dir/d dir2/e f"
+  $ second_files="g dir/nested/h dir/i dir/j dir2/k dir2/l dir/nested/m"
+  $ extra_files="dir/n dir/o p q"
+
+  $ hg init reference-repo
+  $ cd reference-repo
+  $ mkdir -p dir/nested dir2
+  $ touch -t 200001010000 $first_files $directories
+  $ hg commit -Aqm "recreate a bunch of files to facilitate dirstate-v2 append"
+  $ touch -t 200001010010 $second_files $directories
+  $ hg commit -Aqm "more files to have two commits"
+  $ hg log -G -v
+  @  changeset:   1:c349430a1631
+  |  tag:         tip
+  |  user:        test
+  |  date:        Thu Jan 01 00:00:00 1970 +0000
+  |  files:       dir/i dir/j dir/nested/h dir/nested/m dir2/k dir2/l g
+  |  description:
+  |  more files to have two commits
+  |
+  |
+  o  changeset:   0:4f23db756b09
+     user:        test
+     date:        Thu Jan 01 00:00:00 1970 +0000
+     files:       dir/b dir/c dir/d dir/nested/a dir2/e f
+     description:
+     recreate a bunch of files to facilitate dirstate-v2 append
+  
+  
+  $ hg manifest
+  dir/b
+  dir/c
+  dir/d
+  dir/i
+  dir/j
+  dir/nested/a
+  dir/nested/h
+  dir/nested/m
+  dir2/e
+  dir2/k
+  dir2/l
+  f
+  g
+
+Add some unknown files and refresh the dirstate
+
+  $ touch -t 200001010020 $extra_files
+  $ hg add dir/o
+  $ hg remove dir/nested/m
+
+  $ hg st
+  A dir/o
+  R dir/nested/m
+  ? dir/n
+  ? p
+  ? q
+  $ hg debugstate
+  n 644          0 2000-01-01 00:00:00 dir/b
+  n 644          0 2000-01-01 00:00:00 dir/c
+  n 644          0 2000-01-01 00:00:00 dir/d
+  n 644          0 2000-01-01 00:10:00 dir/i
+  n 644          0 2000-01-01 00:10:00 dir/j
+  n 644          0 2000-01-01 00:00:00 dir/nested/a
+  n 644          0 2000-01-01 00:10:00 dir/nested/h
+  r ?????????????????????????????????? dir/nested/m (glob)
+  a ?????????????????????????????????? dir/o (glob)
+  n 644          0 2000-01-01 00:00:00 dir2/e
+  n 644          0 2000-01-01 00:10:00 dir2/k
+  n 644          0 2000-01-01 00:10:00 dir2/l
+  n 644          0 2000-01-01 00:00:00 f
+  n 644          0 2000-01-01 00:10:00 g
+  $ hg debugstate > ../reference
+  $ cd ..
+
+Explain / verify the test principles
+------------------------------------
+
+First, we can properly copy the reference
+
+  $ cp -a reference-repo sanity-check
+  $ cd sanity-check
+  $ hg debugstate
+  n 644          0 2000-01-01 00:00:00 dir/b
+  n 644          0 2000-01-01 00:00:00 dir/c
+  n 644          0 2000-01-01 00:00:00 dir/d
+  n 644          0 2000-01-01 00:10:00 dir/i
+  n 644          0 2000-01-01 00:10:00 dir/j
+  n 644          0 2000-01-01 00:00:00 dir/nested/a
+  n 644          0 2000-01-01 00:10:00 dir/nested/h
+  r ?????????????????????????????????? dir/nested/m (glob)
+  a ?????????????????????????????????? dir/o (glob)
+  n 644          0 2000-01-01 00:00:00 dir2/e
+  n 644          0 2000-01-01 00:10:00 dir2/k
+  n 644          0 2000-01-01 00:10:00 dir2/l
+  n 644          0 2000-01-01 00:00:00 f
+  n 644          0 2000-01-01 00:10:00 g
+  $ hg debugstate > ../post-copy
+  $ diff ../reference ../post-copy
+
+And status thinks the cache is in a proper state
+
+  $ hg st
+  A dir/o
+  R dir/nested/m
+  ? dir/n
+  ? p
+  ? q
+  $ hg debugstate
+  n 644          0 2000-01-01 00:00:00 dir/b
+  n 644          0 2000-01-01 00:00:00 dir/c
+  n 644          0 2000-01-01 00:00:00 dir/d
+  n 644          0 2000-01-01 00:10:00 dir/i
+  n 644          0 2000-01-01 00:10:00 dir/j
+  n 644          0 2000-01-01 00:00:00 dir/nested/a
+  n 644          0 2000-01-01 00:10:00 dir/nested/h
+  r ?????????????????????????????????? dir/nested/m (glob)
+  a ?????????????????????????????????? dir/o (glob)
+  n 644          0 2000-01-01 00:00:00 dir2/e
+  n 644          0 2000-01-01 00:10:00 dir2/k
+  n 644          0 2000-01-01 00:10:00 dir2/l
+  n 644          0 2000-01-01 00:00:00 f
+  n 644          0 2000-01-01 00:10:00 g
+  $ hg debugstate > ../post-status
+  $ diff ../reference ../post-status
+
+Then we can start a status that:
+- has some update to do (the touch call)
+- will wait AFTER running status, but before updating the cache on disk
+
+  $ touch -t 200001010001 dir/c
+  $ hg st >$TESTTMP/status-race-lock.out 2>$TESTTMP/status-race-lock.log \
+  > --config rhg.on-unsupported=abort \
+  > --config devel.sync.status.pre-dirstate-write-file=$TESTTMP/status-race-lock \
+  > &
+  $ $RUNTESTDIR/testlib/wait-on-file 5 $TESTTMP/status-race-lock.waiting
+
+We check it runs the status first by modifying a file and updating another timestamp
+
+  $ touch -t 200001010003 dir/i
+  $ echo babar > dir/j
+  $ touch $TESTTMP/status-race-lock
+  $ wait
+
+The test process should have reported a status before the change we made,
+and should have missed the timestamp update
+
+  $ cat $TESTTMP/status-race-lock.out
+  A dir/o
+  R dir/nested/m
+  ? dir/n
+  ? p
+  ? q
+  $ cat $TESTTMP/status-race-lock.log
+  $ hg debugstate | grep dir/c
+  n 644          0 2000-01-01 00:01:00 dir/c
+  $ hg debugstate | grep dir/i
+  n 644          0 2000-01-01 00:10:00 dir/i
+  $ hg debugstate | grep dir/j
+  n 644          0 2000-01-01 00:10:00 dir/j
+
+final cleanup
+
+  $ rm $TESTTMP/status-race-lock $TESTTMP/status-race-lock.waiting
+  $ cd ..

mercurial/configitems.py

@@ -1,2877 +1,2893 b''
 # configitems.py - centralized declaration of configuration option
 #
 #  Copyright 2017 Pierre-Yves David <[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 functools
 import re
 
 from . import (
     encoding,
     error,
 )
 
 
 def loadconfigtable(ui, extname, configtable):
     """update config item known to the ui with the extension ones"""
     for section, items in sorted(configtable.items()):
         knownitems = ui._knownconfig.setdefault(section, itemregister())
         knownkeys = set(knownitems)
         newkeys = set(items)
         for key in sorted(knownkeys & newkeys):
             msg = b"extension '%s' overwrite config item '%s.%s'"
             msg %= (extname, section, key)
             ui.develwarn(msg, config=b'warn-config')
 
         knownitems.update(items)
 
 
 class configitem:
     """represent a known config item
 
     :section: the official config section where to find this item,
        :name: the official name within the section,
     :default: default value for this item,
     :alias: optional list of tuples as alternatives,
     :generic: this is a generic definition, match name using regular expression.
     """
 
     def __init__(
         self,
         section,
         name,
         default=None,
         alias=(),
         generic=False,
         priority=0,
         experimental=False,
     ):
         self.section = section
         self.name = name
         self.default = default
         self.alias = list(alias)
         self.generic = generic
         self.priority = priority
         self.experimental = experimental
         self._re = None
         if generic:
             self._re = re.compile(self.name)
 
 
 class itemregister(dict):
     """A specialized dictionary that can handle wild-card selection"""
 
     def __init__(self):
         super(itemregister, self).__init__()
         self._generics = set()
 
     def update(self, other):
         super(itemregister, self).update(other)
         self._generics.update(other._generics)
 
     def __setitem__(self, key, item):
         super(itemregister, self).__setitem__(key, item)
         if item.generic:
             self._generics.add(item)
 
     def get(self, key):
         baseitem = super(itemregister, self).get(key)
         if baseitem is not None and not baseitem.generic:
             return baseitem
 
         # search for a matching generic item
         generics = sorted(self._generics, key=(lambda x: (x.priority, x.name)))
         for item in generics:
             # we use 'match' instead of 'search' to make the matching simpler
             # for people unfamiliar with regular expression. Having the match
             # rooted to the start of the string will produce less surprising
             # result for user writing simple regex for sub-attribute.
             #
             # For example using "color\..*" match produces an unsurprising
             # result, while using search could suddenly match apparently
             # unrelated configuration that happens to contains "color."
             # anywhere. This is a tradeoff where we favor requiring ".*" on
             # some match to avoid the need to prefix most pattern with "^".
             # The "^" seems more error prone.
             if item._re.match(key):
                 return item
 
         return None
 
 
 coreitems = {}
 
 
 def _register(configtable, *args, **kwargs):
     item = configitem(*args, **kwargs)
     section = configtable.setdefault(item.section, itemregister())
     if item.name in section:
         msg = b"duplicated config item registration for '%s.%s'"
         raise error.ProgrammingError(msg % (item.section, item.name))
     section[item.name] = item
 
 
 # special value for case where the default is derived from other values
 dynamicdefault = object()
 
 # Registering actual config items
 
 
 def getitemregister(configtable):
     f = functools.partial(_register, configtable)
     # export pseudo enum as configitem.*
     f.dynamicdefault = dynamicdefault
     return f
 
 
 coreconfigitem = getitemregister(coreitems)
 
 
 def _registerdiffopts(section, configprefix=b''):
     coreconfigitem(
         section,
         configprefix + b'nodates',
         default=False,
     )
     coreconfigitem(
         section,
         configprefix + b'showfunc',
         default=False,
     )
     coreconfigitem(
         section,
         configprefix + b'unified',
         default=None,
     )
     coreconfigitem(
         section,
         configprefix + b'git',
         default=False,
     )
     coreconfigitem(
         section,
         configprefix + b'ignorews',
         default=False,
     )
     coreconfigitem(
         section,
         configprefix + b'ignorewsamount',
         default=False,
     )
     coreconfigitem(
         section,
         configprefix + b'ignoreblanklines',
         default=False,
     )
     coreconfigitem(
         section,
         configprefix + b'ignorewseol',
         default=False,
     )
     coreconfigitem(
         section,
         configprefix + b'nobinary',
         default=False,
     )
     coreconfigitem(
         section,
         configprefix + b'noprefix',
         default=False,
     )
     coreconfigitem(
         section,
         configprefix + b'word-diff',
         default=False,
     )
 
 
 coreconfigitem(
     b'alias',
     b'.*',
     default=dynamicdefault,
     generic=True,
 )
 coreconfigitem(
     b'auth',
     b'cookiefile',
     default=None,
 )
 _registerdiffopts(section=b'annotate')
 # bookmarks.pushing: internal hack for discovery
 coreconfigitem(
     b'bookmarks',
     b'pushing',
     default=list,
 )
 # bundle.mainreporoot: internal hack for bundlerepo
 coreconfigitem(
     b'bundle',
     b'mainreporoot',
     default=b'',
 )
 coreconfigitem(
     b'censor',
     b'policy',
     default=b'abort',
     experimental=True,
 )
 coreconfigitem(
     b'chgserver',
     b'idletimeout',
     default=3600,
 )
 coreconfigitem(
     b'chgserver',
     b'skiphash',
     default=False,
 )
 coreconfigitem(
     b'cmdserver',
     b'log',
     default=None,
 )
 coreconfigitem(
     b'cmdserver',
     b'max-log-files',
     default=7,
 )
 coreconfigitem(
     b'cmdserver',
     b'max-log-size',
     default=b'1 MB',
 )
 coreconfigitem(
     b'cmdserver',
     b'max-repo-cache',
     default=0,
     experimental=True,
 )
 coreconfigitem(
     b'cmdserver',
     b'message-encodings',
     default=list,
 )
 coreconfigitem(
     b'cmdserver',
     b'track-log',
     default=lambda: [b'chgserver', b'cmdserver', b'repocache'],
 )
 coreconfigitem(
     b'cmdserver',
     b'shutdown-on-interrupt',
     default=True,
 )
 coreconfigitem(
     b'color',
     b'.*',
     default=None,
     generic=True,
 )
 coreconfigitem(
     b'color',
     b'mode',
     default=b'auto',
 )
 coreconfigitem(
     b'color',
     b'pagermode',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'command-templates',
     b'graphnode',
     default=None,
     alias=[(b'ui', b'graphnodetemplate')],
 )
 coreconfigitem(
     b'command-templates',
     b'log',
     default=None,
     alias=[(b'ui', b'logtemplate')],
 )
 coreconfigitem(
     b'command-templates',
     b'mergemarker',
     default=(
         b'{node|short} '
         b'{ifeq(tags, "tip", "", '
         b'ifeq(tags, "", "", "{tags} "))}'
         b'{if(bookmarks, "{bookmarks} ")}'
         b'{ifeq(branch, "default", "", "{branch} ")}'
         b'- {author|user}: {desc|firstline}'
     ),
     alias=[(b'ui', b'mergemarkertemplate')],
 )
 coreconfigitem(
     b'command-templates',
     b'pre-merge-tool-output',
     default=None,
     alias=[(b'ui', b'pre-merge-tool-output-template')],
 )
 coreconfigitem(
     b'command-templates',
     b'oneline-summary',
     default=None,
 )
 coreconfigitem(
     b'command-templates',
     b'oneline-summary.*',
     default=dynamicdefault,
     generic=True,
 )
 _registerdiffopts(section=b'commands', configprefix=b'commit.interactive.')
 coreconfigitem(
     b'commands',
     b'commit.post-status',
     default=False,
 )
 coreconfigitem(
     b'commands',
     b'grep.all-files',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'commands',
     b'merge.require-rev',
     default=False,
 )
 coreconfigitem(
     b'commands',
     b'push.require-revs',
     default=False,
 )
 coreconfigitem(
     b'commands',
     b'resolve.confirm',
     default=False,
 )
 coreconfigitem(
     b'commands',
     b'resolve.explicit-re-merge',
     default=False,
 )
 coreconfigitem(
     b'commands',
     b'resolve.mark-check',
     default=b'none',
 )
 _registerdiffopts(section=b'commands', configprefix=b'revert.interactive.')
 coreconfigitem(
     b'commands',
     b'show.aliasprefix',
     default=list,
 )
 coreconfigitem(
     b'commands',
     b'status.relative',
     default=False,
 )
 coreconfigitem(
     b'commands',
     b'status.skipstates',
     default=[],
     experimental=True,
 )
 coreconfigitem(
     b'commands',
     b'status.terse',
     default=b'',
 )
 coreconfigitem(
     b'commands',
     b'status.verbose',
     default=False,
 )
 coreconfigitem(
     b'commands',
     b'update.check',
     default=None,
 )
 coreconfigitem(
     b'commands',
     b'update.requiredest',
     default=False,
 )
 coreconfigitem(
     b'committemplate',
     b'.*',
     default=None,
     generic=True,
 )
 coreconfigitem(
     b'convert',
     b'bzr.saverev',
     default=True,
 )
 coreconfigitem(
     b'convert',
     b'cvsps.cache',
     default=True,
 )
 coreconfigitem(
     b'convert',
     b'cvsps.fuzz',
     default=60,
 )
 coreconfigitem(
     b'convert',
     b'cvsps.logencoding',
     default=None,
 )
 coreconfigitem(
     b'convert',
     b'cvsps.mergefrom',
     default=None,
 )
 coreconfigitem(
     b'convert',
     b'cvsps.mergeto',
     default=None,
 )
 coreconfigitem(
     b'convert',
     b'git.committeractions',
     default=lambda: [b'messagedifferent'],
 )
 coreconfigitem(
     b'convert',
     b'git.extrakeys',
     default=list,
 )
 coreconfigitem(
     b'convert',
     b'git.findcopiesharder',
     default=False,
 )
 coreconfigitem(
     b'convert',
     b'git.remoteprefix',
     default=b'remote',
 )
 coreconfigitem(
     b'convert',
     b'git.renamelimit',
     default=400,
 )
 coreconfigitem(
     b'convert',
     b'git.saverev',
     default=True,
 )
 coreconfigitem(
     b'convert',
     b'git.similarity',
     default=50,
 )
 coreconfigitem(
     b'convert',
     b'git.skipsubmodules',
     default=False,
 )
 coreconfigitem(
     b'convert',
     b'hg.clonebranches',
     default=False,
 )
 coreconfigitem(
     b'convert',
     b'hg.ignoreerrors',
     default=False,
 )
 coreconfigitem(
     b'convert',
     b'hg.preserve-hash',
     default=False,
 )
 coreconfigitem(
     b'convert',
     b'hg.revs',
     default=None,
 )
 coreconfigitem(
     b'convert',
     b'hg.saverev',
     default=False,
 )
 coreconfigitem(
     b'convert',
     b'hg.sourcename',
     default=None,
 )
 coreconfigitem(
     b'convert',
     b'hg.startrev',
     default=None,
 )
 coreconfigitem(
     b'convert',
     b'hg.tagsbranch',
     default=b'default',
 )
 coreconfigitem(
     b'convert',
     b'hg.usebranchnames',
     default=True,
 )
 coreconfigitem(
     b'convert',
     b'ignoreancestorcheck',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'convert',
     b'localtimezone',
     default=False,
 )
 coreconfigitem(
     b'convert',
     b'p4.encoding',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'convert',
     b'p4.startrev',
     default=0,
 )
 coreconfigitem(
     b'convert',
     b'skiptags',
     default=False,
 )
 coreconfigitem(
     b'convert',
     b'svn.debugsvnlog',
     default=True,
 )
 coreconfigitem(
     b'convert',
     b'svn.trunk',
     default=None,
 )
 coreconfigitem(
     b'convert',
     b'svn.tags',
     default=None,
 )
 coreconfigitem(
     b'convert',
     b'svn.branches',
     default=None,
 )
 coreconfigitem(
     b'convert',
     b'svn.startrev',
     default=0,
 )
 coreconfigitem(
     b'convert',
     b'svn.dangerous-set-commit-dates',
     default=False,
 )
 coreconfigitem(
     b'debug',
     b'dirstate.delaywrite',
     default=0,
 )
 coreconfigitem(
     b'debug',
     b'revlog.verifyposition.changelog',
     default=b'',
 )
 coreconfigitem(
     b'debug',
     b'revlog.debug-delta',
     default=False,
 )
 coreconfigitem(
     b'defaults',
     b'.*',
     default=None,
     generic=True,
 )
 coreconfigitem(
     b'devel',
     b'all-warnings',
     default=False,
 )
 coreconfigitem(
     b'devel',
     b'bundle2.debug',
     default=False,
 )
 coreconfigitem(
     b'devel',
     b'bundle.delta',
     default=b'',
 )
 coreconfigitem(
     b'devel',
     b'cache-vfs',
     default=None,
 )
 coreconfigitem(
     b'devel',
     b'check-locks',
     default=False,
 )
 coreconfigitem(
     b'devel',
     b'check-relroot',
     default=False,
 )
 # Track copy information for all file, not just "added" one (very slow)
 coreconfigitem(
     b'devel',
     b'copy-tracing.trace-all-files',
     default=False,
 )
 coreconfigitem(
     b'devel',
     b'default-date',
     default=None,
 )
 coreconfigitem(
     b'devel',
     b'deprec-warn',
     default=False,
 )
 coreconfigitem(
     b'devel',
     b'disableloaddefaultcerts',
     default=False,
 )
 coreconfigitem(
     b'devel',
     b'warn-empty-changegroup',
     default=False,
 )
 coreconfigitem(
     b'devel',
     b'legacy.exchange',
     default=list,
 )
 # When True, revlogs use a special reference version of the nodemap, that is not
 # performant but is "known" to behave properly.
 coreconfigitem(
     b'devel',
     b'persistent-nodemap',
     default=False,
 )
 coreconfigitem(
     b'devel',
     b'servercafile',
     default=b'',
 )
 coreconfigitem(
     b'devel',
     b'serverexactprotocol',
     default=b'',
 )
 coreconfigitem(
     b'devel',
     b'serverrequirecert',
     default=False,
 )
+# Makes the status algorithm wait for the existence of this file
+# (or until a timeout of `devel.sync.status.pre-dirstate-write-file-timeout`
+# seconds) before taking the lock and writing the dirstate.
+# Status signals that it's ready to wait by creating a file
+# with the same name + `.waiting`.
+# Useful when testing race conditions.
+coreconfigitem(
+    b'devel',
+    b'sync.status.pre-dirstate-write-file',
+    default=None,
+)
+coreconfigitem(
+    b'devel',
+    b'sync.status.pre-dirstate-write-file-timeout',
+    default=2,
+)
 coreconfigitem(
     b'devel',
     b'strip-obsmarkers',
     default=True,
 )
 coreconfigitem(
     b'devel',
     b'warn-config',
     default=None,
 )
 coreconfigitem(
     b'devel',
     b'warn-config-default',
     default=None,
 )
 coreconfigitem(
     b'devel',
     b'user.obsmarker',
     default=None,
 )
 coreconfigitem(
     b'devel',
     b'warn-config-unknown',
     default=None,
 )
 coreconfigitem(
     b'devel',
     b'debug.copies',
     default=False,
 )
 coreconfigitem(
     b'devel',
     b'copy-tracing.multi-thread',
     default=True,
 )
 coreconfigitem(
     b'devel',
     b'debug.extensions',
     default=False,
 )
 coreconfigitem(
     b'devel',
     b'debug.repo-filters',
     default=False,
 )
 coreconfigitem(
     b'devel',
     b'debug.peer-request',
     default=False,
 )
 # If discovery.exchange-heads is False, the discovery will not start with
 # remote head fetching and local head querying.
 coreconfigitem(
     b'devel',
     b'discovery.exchange-heads',
     default=True,
 )
 # If discovery.grow-sample is False, the sample size used in set discovery will
 # not be increased through the process
 coreconfigitem(
     b'devel',
     b'discovery.grow-sample',
     default=True,
 )
 # When discovery.grow-sample.dynamic is True, the default, the sample size is
 # adapted to the shape of the undecided set (it is set to the max of:
 # <target-size>, len(roots(undecided)), len(heads(undecided)
 coreconfigitem(
     b'devel',
     b'discovery.grow-sample.dynamic',
     default=True,
 )
 # discovery.grow-sample.rate control the rate at which the sample grow
 coreconfigitem(
     b'devel',
     b'discovery.grow-sample.rate',
     default=1.05,
 )
 # If discovery.randomize is False, random sampling during discovery are
 # deterministic. It is meant for integration tests.
 coreconfigitem(
     b'devel',
     b'discovery.randomize',
     default=True,
 )
 # Control the initial size of the discovery sample
 coreconfigitem(
     b'devel',
     b'discovery.sample-size',
     default=200,
 )
 # Control the initial size of the discovery for initial change
 coreconfigitem(
     b'devel',
     b'discovery.sample-size.initial',
     default=100,
 )
 _registerdiffopts(section=b'diff')
 coreconfigitem(
     b'diff',
     b'merge',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'email',
     b'bcc',
     default=None,
 )
 coreconfigitem(
     b'email',
     b'cc',
     default=None,
 )
 coreconfigitem(
     b'email',
     b'charsets',
     default=list,
 )
 coreconfigitem(
     b'email',
     b'from',
     default=None,
 )
 coreconfigitem(
     b'email',
     b'method',
     default=b'smtp',
 )
 coreconfigitem(
     b'email',
     b'reply-to',
     default=None,
 )
 coreconfigitem(
     b'email',
     b'to',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'archivemetatemplate',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'experimental',
     b'auto-publish',
     default=b'publish',
 )
 coreconfigitem(
     b'experimental',
     b'bundle-phases',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'bundle2-advertise',
     default=True,
 )
 coreconfigitem(
     b'experimental',
     b'bundle2-output-capture',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'bundle2.pushback',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'bundle2lazylocking',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'bundlecomplevel',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'bundlecomplevel.bzip2',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'bundlecomplevel.gzip',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'bundlecomplevel.none',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'bundlecomplevel.zstd',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'bundlecompthreads',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'bundlecompthreads.bzip2',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'bundlecompthreads.gzip',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'bundlecompthreads.none',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'bundlecompthreads.zstd',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'changegroup3',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'changegroup4',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'cleanup-as-archived',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'clientcompressionengines',
     default=list,
 )
 coreconfigitem(
     b'experimental',
     b'copytrace',
     default=b'on',
 )
 coreconfigitem(
     b'experimental',
     b'copytrace.movecandidateslimit',
     default=100,
 )
 coreconfigitem(
     b'experimental',
     b'copytrace.sourcecommitlimit',
     default=100,
 )
 coreconfigitem(
     b'experimental',
     b'copies.read-from',
     default=b"filelog-only",
 )
 coreconfigitem(
     b'experimental',
     b'copies.write-to',
     default=b'filelog-only',
 )
 coreconfigitem(
     b'experimental',
     b'crecordtest',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'directaccess',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'directaccess.revnums',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'editortmpinhg',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'evolution',
     default=list,
 )
 coreconfigitem(
     b'experimental',
     b'evolution.allowdivergence',
     default=False,
     alias=[(b'experimental', b'allowdivergence')],
 )
 coreconfigitem(
     b'experimental',
     b'evolution.allowunstable',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'evolution.createmarkers',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'evolution.effect-flags',
     default=True,
     alias=[(b'experimental', b'effect-flags')],
 )
 coreconfigitem(
     b'experimental',
     b'evolution.exchange',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'evolution.bundle-obsmarker',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'evolution.bundle-obsmarker:mandatory',
     default=True,
 )
 coreconfigitem(
     b'experimental',
     b'log.topo',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'evolution.report-instabilities',
     default=True,
 )
 coreconfigitem(
     b'experimental',
     b'evolution.track-operation',
     default=True,
 )
 # repo-level config to exclude a revset visibility
 #
 # The target use case is to use `share` to expose different subset of the same
 # repository, especially server side. See also `server.view`.
 coreconfigitem(
     b'experimental',
     b'extra-filter-revs',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'maxdeltachainspan',
     default=-1,
 )
 # tracks files which were undeleted (merge might delete them but we explicitly
 # kept/undeleted them) and creates new filenodes for them
 coreconfigitem(
     b'experimental',
     b'merge-track-salvaged',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'mmapindexthreshold',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'narrow',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'nonnormalparanoidcheck',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'exportableenviron',
     default=list,
 )
 coreconfigitem(
     b'experimental',
     b'extendedheader.index',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'extendedheader.similarity',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'graphshorten',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'graphstyle.parent',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'experimental',
     b'graphstyle.missing',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'experimental',
     b'graphstyle.grandparent',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'experimental',
     b'hook-track-tags',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'httppostargs',
     default=False,
 )
 coreconfigitem(b'experimental', b'nointerrupt', default=False)
 coreconfigitem(b'experimental', b'nointerrupt-interactiveonly', default=True)
 
 coreconfigitem(
     b'experimental',
     b'obsmarkers-exchange-debug',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'remotenames',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'removeemptydirs',
     default=True,
 )
 coreconfigitem(
     b'experimental',
     b'revert.interactive.select-to-keep',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'revisions.prefixhexnode',
     default=False,
 )
 # "out of experimental" todo list.
 #
 # * include management of a persistent nodemap in the main docket
 # * enforce a "no-truncate" policy for mmap safety
 #      - for censoring operation
 #      - for stripping operation
 #      - for rollback operation
 # * proper streaming (race free) of the docket file
 # * track garbage data to evemtually allow rewriting -existing- sidedata.
 # * Exchange-wise, we will also need to do something more efficient than
 #   keeping references to the affected revlogs, especially memory-wise when
 #   rewriting sidedata.
 # * introduce a proper solution to reduce the number of filelog related files.
 # * use caching for reading sidedata (similar to what we do for data).
 # * no longer set offset=0 if sidedata_size=0 (simplify cutoff computation).
 # * Improvement to consider
 #   - avoid compression header in chunk using the default compression?
 #   - forbid "inline" compression mode entirely?
 #   - split the data offset and flag field (the 2 bytes save are mostly trouble)
 #   - keep track of uncompressed -chunk- size (to preallocate memory better)
 #   - keep track of chain base or size (probably not that useful anymore)
 coreconfigitem(
     b'experimental',
     b'revlogv2',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'revisions.disambiguatewithin',
     default=None,
 )
 coreconfigitem(
     b'experimental',
     b'rust.index',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'server.filesdata.recommended-batch-size',
     default=50000,
 )
 coreconfigitem(
     b'experimental',
     b'server.manifestdata.recommended-batch-size',
     default=100000,
 )
 coreconfigitem(
     b'experimental',
     b'server.stream-narrow-clones',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'single-head-per-branch',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'single-head-per-branch:account-closed-heads',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'single-head-per-branch:public-changes-only',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'sparse-read',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'sparse-read.density-threshold',
     default=0.50,
 )
 coreconfigitem(
     b'experimental',
     b'sparse-read.min-gap-size',
     default=b'65K',
 )
 coreconfigitem(
     b'experimental',
     b'treemanifest',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'update.atomic-file',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'web.full-garbage-collection-rate',
     default=1,  # still forcing a full collection on each request
 )
 coreconfigitem(
     b'experimental',
     b'worker.wdir-get-thread-safe',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'worker.repository-upgrade',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'xdiff',
     default=False,
 )
 coreconfigitem(
     b'extensions',
     b'[^:]*',
     default=None,
     generic=True,
 )
 coreconfigitem(
     b'extensions',
     b'[^:]*:required',
     default=False,
     generic=True,
 )
 coreconfigitem(
     b'extdata',
     b'.*',
     default=None,
     generic=True,
 )
 coreconfigitem(
     b'format',
     b'bookmarks-in-store',
     default=False,
 )
 coreconfigitem(
     b'format',
     b'chunkcachesize',
     default=None,
     experimental=True,
 )
 coreconfigitem(
     # Enable this dirstate format *when creating a new repository*.
     # Which format to use for existing repos is controlled by .hg/requires
     b'format',
     b'use-dirstate-v2',
     default=False,
     experimental=True,
     alias=[(b'format', b'exp-rc-dirstate-v2')],
 )
 coreconfigitem(
     b'format',
     b'use-dirstate-v2.automatic-upgrade-of-mismatching-repositories',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'format',
     b'use-dirstate-v2.automatic-upgrade-of-mismatching-repositories:quiet',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'format',
     b'use-dirstate-tracked-hint',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'format',
     b'use-dirstate-tracked-hint.version',
     default=1,
     experimental=True,
 )
 coreconfigitem(
     b'format',
     b'use-dirstate-tracked-hint.automatic-upgrade-of-mismatching-repositories',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'format',
     b'use-dirstate-tracked-hint.automatic-upgrade-of-mismatching-repositories:quiet',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'format',
     b'dotencode',
     default=True,
 )
 coreconfigitem(
     b'format',
     b'generaldelta',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'format',
     b'manifestcachesize',
     default=None,
     experimental=True,
 )
 coreconfigitem(
     b'format',
     b'maxchainlen',
     default=dynamicdefault,
     experimental=True,
 )
 coreconfigitem(
     b'format',
     b'obsstore-version',
     default=None,
 )
 coreconfigitem(
     b'format',
     b'sparse-revlog',
     default=True,
 )
 coreconfigitem(
     b'format',
     b'revlog-compression',
     default=lambda: [b'zstd', b'zlib'],
     alias=[(b'experimental', b'format.compression')],
 )
 # Experimental TODOs:
 #
 # * Same as for revlogv2 (but for the reduction of the number of files)
 # * Actually computing the rank of changesets
 # * Improvement to investigate
 #   - storing .hgtags fnode
 #   - storing branch related identifier
 
 coreconfigitem(
     b'format',
     b'exp-use-changelog-v2',
     default=None,
     experimental=True,
 )
 coreconfigitem(
     b'format',
     b'usefncache',
     default=True,
 )
 coreconfigitem(
     b'format',
     b'usegeneraldelta',
     default=True,
 )
 coreconfigitem(
     b'format',
     b'usestore',
     default=True,
 )
 
 
 def _persistent_nodemap_default():
     """compute `use-persistent-nodemap` default value
 
     The feature is disabled unless a fast implementation is available.
     """
     from . import policy
 
     return policy.importrust('revlog') is not None
 
 
 coreconfigitem(
     b'format',
     b'use-persistent-nodemap',
     default=_persistent_nodemap_default,
 )
 coreconfigitem(
     b'format',
     b'exp-use-copies-side-data-changeset',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'format',
     b'use-share-safe',
     default=True,
 )
 coreconfigitem(
     b'format',
     b'use-share-safe.automatic-upgrade-of-mismatching-repositories',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'format',
     b'use-share-safe.automatic-upgrade-of-mismatching-repositories:quiet',
     default=False,
     experimental=True,
 )
 
 # Moving this on by default means we are confident about the scaling of phases.
 # This is not garanteed to be the case at the time this message is written.
 coreconfigitem(
     b'format',
     b'use-internal-phase',
     default=False,
     experimental=True,
 )
 # The interaction between the archived phase and obsolescence markers needs to
 # be sorted out before wider usage of this are to be considered.
 #
 # At the time this message is written, behavior when archiving obsolete
 # changeset differ significantly from stripping. As part of stripping, we also
 # remove the obsolescence marker associated to the stripped changesets,
 # revealing the precedecessors changesets when applicable. When archiving, we
 # don't touch the obsolescence markers, keeping everything hidden. This can
 # result in quite confusing situation for people combining exchanging draft
 # with the archived phases. As some markers needed by others may be skipped
 # during exchange.
 coreconfigitem(
     b'format',
     b'exp-archived-phase',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'shelve',
     b'store',
     default=b'internal',
     experimental=True,
 )
 coreconfigitem(
     b'fsmonitor',
     b'warn_when_unused',
     default=True,
 )
 coreconfigitem(
     b'fsmonitor',
     b'warn_update_file_count',
     default=50000,
 )
 coreconfigitem(
     b'fsmonitor',
     b'warn_update_file_count_rust',
     default=400000,
 )
 coreconfigitem(
     b'help',
     br'hidden-command\..*',
     default=False,
     generic=True,
 )
 coreconfigitem(
     b'help',
     br'hidden-topic\..*',
     default=False,
     generic=True,
 )
 coreconfigitem(
     b'hooks',
     b'[^:]*',
     default=dynamicdefault,
     generic=True,
 )
 coreconfigitem(
     b'hooks',
     b'.*:run-with-plain',
     default=True,
     generic=True,
 )
 coreconfigitem(
     b'hgweb-paths',
     b'.*',
     default=list,
     generic=True,
 )
 coreconfigitem(
     b'hostfingerprints',
     b'.*',
     default=list,
     generic=True,
 )
 coreconfigitem(
     b'hostsecurity',
     b'ciphers',
     default=None,
 )
 coreconfigitem(
     b'hostsecurity',
     b'minimumprotocol',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'hostsecurity',
     b'.*:minimumprotocol$',
     default=dynamicdefault,
     generic=True,
 )
 coreconfigitem(
     b'hostsecurity',
     b'.*:ciphers$',
     default=dynamicdefault,
     generic=True,
 )
 coreconfigitem(
     b'hostsecurity',
     b'.*:fingerprints$',
     default=list,
     generic=True,
 )
 coreconfigitem(
     b'hostsecurity',
     b'.*:verifycertsfile$',
     default=None,
     generic=True,
 )
 
 coreconfigitem(
     b'http_proxy',
     b'always',
     default=False,
 )
 coreconfigitem(
     b'http_proxy',
     b'host',
     default=None,
 )
 coreconfigitem(
     b'http_proxy',
     b'no',
     default=list,
 )
 coreconfigitem(
     b'http_proxy',
     b'passwd',
     default=None,
 )
 coreconfigitem(
     b'http_proxy',
     b'user',
     default=None,
 )
 
 coreconfigitem(
     b'http',
     b'timeout',
     default=None,
 )
 
 coreconfigitem(
     b'logtoprocess',
     b'commandexception',
     default=None,
 )
 coreconfigitem(
     b'logtoprocess',
     b'commandfinish',
     default=None,
 )
 coreconfigitem(
     b'logtoprocess',
     b'command',
     default=None,
 )
 coreconfigitem(
     b'logtoprocess',
     b'develwarn',
     default=None,
 )
 coreconfigitem(
     b'logtoprocess',
     b'uiblocked',
     default=None,
 )
 coreconfigitem(
     b'merge',
     b'checkunknown',
     default=b'abort',
 )
 coreconfigitem(
     b'merge',
     b'checkignored',
     default=b'abort',
 )
 coreconfigitem(
     b'experimental',
     b'merge.checkpathconflicts',
     default=False,
 )
 coreconfigitem(
     b'merge',
     b'followcopies',
     default=True,
 )
 coreconfigitem(
     b'merge',
     b'on-failure',
     default=b'continue',
 )
 coreconfigitem(
     b'merge',
     b'preferancestor',
     default=lambda: [b'*'],
     experimental=True,
 )
 coreconfigitem(
     b'merge',
     b'strict-capability-check',
     default=False,
 )
 coreconfigitem(
     b'merge',
     b'disable-partial-tools',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'partial-merge-tools',
     b'.*',
     default=None,
     generic=True,
     experimental=True,
 )
 coreconfigitem(
     b'partial-merge-tools',
     br'.*\.patterns',
     default=dynamicdefault,
     generic=True,
     priority=-1,
     experimental=True,
 )
 coreconfigitem(
     b'partial-merge-tools',
     br'.*\.executable$',
     default=dynamicdefault,
     generic=True,
     priority=-1,
     experimental=True,
 )
 coreconfigitem(
     b'partial-merge-tools',
     br'.*\.order',
     default=0,
     generic=True,
     priority=-1,
     experimental=True,
 )
 coreconfigitem(
     b'partial-merge-tools',
     br'.*\.args',
     default=b"$local $base $other",
     generic=True,
     priority=-1,
     experimental=True,
 )
 coreconfigitem(
     b'partial-merge-tools',
     br'.*\.disable',
     default=False,
     generic=True,
     priority=-1,
     experimental=True,
 )
 coreconfigitem(
     b'merge-tools',
     b'.*',
     default=None,
     generic=True,
 )
 coreconfigitem(
     b'merge-tools',
     br'.*\.args$',
     default=b"$local $base $other",
     generic=True,
     priority=-1,
 )
 coreconfigitem(
     b'merge-tools',
     br'.*\.binary$',
     default=False,
     generic=True,
     priority=-1,
 )
 coreconfigitem(
     b'merge-tools',
     br'.*\.check$',
     default=list,
     generic=True,
     priority=-1,
 )
 coreconfigitem(
     b'merge-tools',
     br'.*\.checkchanged$',
     default=False,
     generic=True,
     priority=-1,
 )
 coreconfigitem(
     b'merge-tools',
     br'.*\.executable$',
     default=dynamicdefault,
     generic=True,
     priority=-1,
 )
 coreconfigitem(
     b'merge-tools',
     br'.*\.fixeol$',
     default=False,
     generic=True,
     priority=-1,
 )
 coreconfigitem(
     b'merge-tools',
     br'.*\.gui$',
     default=False,
     generic=True,
     priority=-1,
 )
 coreconfigitem(
     b'merge-tools',
     br'.*\.mergemarkers$',
     default=b'basic',
     generic=True,
     priority=-1,
 )
 coreconfigitem(
     b'merge-tools',
     br'.*\.mergemarkertemplate$',
     default=dynamicdefault,  # take from command-templates.mergemarker
     generic=True,
     priority=-1,
 )
 coreconfigitem(
     b'merge-tools',
     br'.*\.priority$',
     default=0,
     generic=True,
     priority=-1,
 )
 coreconfigitem(
     b'merge-tools',
     br'.*\.premerge$',
     default=dynamicdefault,
     generic=True,
     priority=-1,
 )
 coreconfigitem(
     b'merge-tools',
     br'.*\.symlink$',
     default=False,
     generic=True,
     priority=-1,
 )
 coreconfigitem(
     b'pager',
     b'attend-.*',
     default=dynamicdefault,
     generic=True,
 )
 coreconfigitem(
     b'pager',
     b'ignore',
     default=list,
 )
 coreconfigitem(
     b'pager',
     b'pager',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'patch',
     b'eol',
     default=b'strict',
 )
 coreconfigitem(
     b'patch',
     b'fuzz',
     default=2,
 )
 coreconfigitem(
     b'paths',
     b'default',
     default=None,
 )
 coreconfigitem(
     b'paths',
     b'default-push',
     default=None,
 )
 coreconfigitem(
     b'paths',
     b'.*',
     default=None,
     generic=True,
 )
 coreconfigitem(
     b'paths',
     b'.*:bookmarks.mode',
     default='default',
     generic=True,
 )
 coreconfigitem(
     b'paths',
     b'.*:multi-urls',
     default=False,
     generic=True,
 )
 coreconfigitem(
     b'paths',
     b'.*:pushrev',
     default=None,
     generic=True,
 )
 coreconfigitem(
     b'paths',
     b'.*:pushurl',
     default=None,
     generic=True,
 )
 coreconfigitem(
     b'phases',
     b'checksubrepos',
     default=b'follow',
 )
 coreconfigitem(
     b'phases',
     b'new-commit',
     default=b'draft',
 )
 coreconfigitem(
     b'phases',
     b'publish',
     default=True,
 )
 coreconfigitem(
     b'profiling',
     b'enabled',
     default=False,
 )
 coreconfigitem(
     b'profiling',
     b'format',
     default=b'text',
 )
 coreconfigitem(
     b'profiling',
     b'freq',
     default=1000,
 )
 coreconfigitem(
     b'profiling',
     b'limit',
     default=30,
 )
 coreconfigitem(
     b'profiling',
     b'nested',
     default=0,
 )
 coreconfigitem(
     b'profiling',
     b'output',
     default=None,
 )
 coreconfigitem(
     b'profiling',
     b'showmax',
     default=0.999,
 )
 coreconfigitem(
     b'profiling',
     b'showmin',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'profiling',
     b'showtime',
     default=True,
 )
 coreconfigitem(
     b'profiling',
     b'sort',
     default=b'inlinetime',
 )
 coreconfigitem(
     b'profiling',
     b'statformat',
     default=b'hotpath',
 )
 coreconfigitem(
     b'profiling',
     b'time-track',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'profiling',
     b'type',
     default=b'stat',
 )
 coreconfigitem(
     b'progress',
     b'assume-tty',
     default=False,
 )
 coreconfigitem(
     b'progress',
     b'changedelay',
     default=1,
 )
 coreconfigitem(
     b'progress',
     b'clear-complete',
     default=True,
 )
 coreconfigitem(
     b'progress',
     b'debug',
     default=False,
 )
 coreconfigitem(
     b'progress',
     b'delay',
     default=3,
 )
 coreconfigitem(
     b'progress',
     b'disable',
     default=False,
 )
 coreconfigitem(
     b'progress',
     b'estimateinterval',
     default=60.0,
 )
 coreconfigitem(
     b'progress',
     b'format',
     default=lambda: [b'topic', b'bar', b'number', b'estimate'],
 )
 coreconfigitem(
     b'progress',
     b'refresh',
     default=0.1,
 )
 coreconfigitem(
     b'progress',
     b'width',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'pull',
     b'confirm',
     default=False,
 )
 coreconfigitem(
     b'push',
     b'pushvars.server',
     default=False,
 )
 coreconfigitem(
     b'rewrite',
     b'backup-bundle',
     default=True,
     alias=[(b'ui', b'history-editing-backup')],
 )
 coreconfigitem(
     b'rewrite',
     b'update-timestamp',
     default=False,
 )
 coreconfigitem(
     b'rewrite',
     b'empty-successor',
     default=b'skip',
     experimental=True,
 )
 # experimental as long as format.use-dirstate-v2 is.
 coreconfigitem(
     b'storage',
     b'dirstate-v2.slow-path',
     default=b"abort",
     experimental=True,
 )
 coreconfigitem(
     b'storage',
     b'new-repo-backend',
     default=b'revlogv1',
     experimental=True,
 )
 coreconfigitem(
     b'storage',
     b'revlog.optimize-delta-parent-choice',
     default=True,
     alias=[(b'format', b'aggressivemergedeltas')],
 )
 coreconfigitem(
     b'storage',
     b'revlog.issue6528.fix-incoming',
     default=True,
 )
 # experimental as long as rust is experimental (or a C version is implemented)
 coreconfigitem(
     b'storage',
     b'revlog.persistent-nodemap.mmap',
     default=True,
 )
 # experimental as long as format.use-persistent-nodemap is.
 coreconfigitem(
     b'storage',
     b'revlog.persistent-nodemap.slow-path',
     default=b"abort",
 )
 
 coreconfigitem(
     b'storage',
     b'revlog.reuse-external-delta',
     default=True,
 )
 coreconfigitem(
     b'storage',
     b'revlog.reuse-external-delta-parent',
     default=None,
 )
 coreconfigitem(
     b'storage',
     b'revlog.zlib.level',
     default=None,
 )
 coreconfigitem(
     b'storage',
     b'revlog.zstd.level',
     default=None,
 )
 coreconfigitem(
     b'server',
     b'bookmarks-pushkey-compat',
     default=True,
 )
 coreconfigitem(
     b'server',
     b'bundle1',
     default=True,
 )
 coreconfigitem(
     b'server',
     b'bundle1gd',
     default=None,
 )
 coreconfigitem(
     b'server',
     b'bundle1.pull',
     default=None,
 )
 coreconfigitem(
     b'server',
     b'bundle1gd.pull',
     default=None,
 )
 coreconfigitem(
     b'server',
     b'bundle1.push',
     default=None,
 )
 coreconfigitem(
     b'server',
     b'bundle1gd.push',
     default=None,
 )
 coreconfigitem(
     b'server',
     b'bundle2.stream',
     default=True,
     alias=[(b'experimental', b'bundle2.stream')],
 )
 coreconfigitem(
     b'server',
     b'compressionengines',
     default=list,
 )
 coreconfigitem(
     b'server',
     b'concurrent-push-mode',
     default=b'check-related',
 )
 coreconfigitem(
     b'server',
     b'disablefullbundle',
     default=False,
 )
 coreconfigitem(
     b'server',
     b'maxhttpheaderlen',
     default=1024,
 )
 coreconfigitem(
     b'server',
     b'pullbundle',
     default=False,
 )
 coreconfigitem(
     b'server',
     b'preferuncompressed',
     default=False,
 )
 coreconfigitem(
     b'server',
     b'streamunbundle',
     default=False,
 )
 coreconfigitem(
     b'server',
     b'uncompressed',
     default=True,
 )
 coreconfigitem(
     b'server',
     b'uncompressedallowsecret',
     default=False,
 )
 coreconfigitem(
     b'server',
     b'view',
     default=b'served',
 )
 coreconfigitem(
     b'server',
     b'validate',
     default=False,
 )
 coreconfigitem(
     b'server',
     b'zliblevel',
     default=-1,
 )
 coreconfigitem(
     b'server',
     b'zstdlevel',
     default=3,
 )
 coreconfigitem(
     b'share',
     b'pool',
     default=None,
 )
 coreconfigitem(
     b'share',
     b'poolnaming',
     default=b'identity',
 )
 coreconfigitem(
     b'share',
     b'safe-mismatch.source-not-safe',
     default=b'abort',
 )
 coreconfigitem(
     b'share',
     b'safe-mismatch.source-safe',
     default=b'abort',
 )
 coreconfigitem(
     b'share',
     b'safe-mismatch.source-not-safe.warn',
     default=True,
 )
 coreconfigitem(
     b'share',
     b'safe-mismatch.source-safe.warn',
     default=True,
 )
 coreconfigitem(
     b'share',
     b'safe-mismatch.source-not-safe:verbose-upgrade',
     default=True,
 )
 coreconfigitem(
     b'share',
     b'safe-mismatch.source-safe:verbose-upgrade',
     default=True,
 )
 coreconfigitem(
     b'shelve',
     b'maxbackups',
     default=10,
 )
 coreconfigitem(
     b'smtp',
     b'host',
     default=None,
 )
 coreconfigitem(
     b'smtp',
     b'local_hostname',
     default=None,
 )
 coreconfigitem(
     b'smtp',
     b'password',
     default=None,
 )
 coreconfigitem(
     b'smtp',
     b'port',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'smtp',
     b'tls',
     default=b'none',
 )
 coreconfigitem(
     b'smtp',
     b'username',
     default=None,
 )
 coreconfigitem(
     b'sparse',
     b'missingwarning',
     default=True,
     experimental=True,
 )
 coreconfigitem(
     b'subrepos',
     b'allowed',
     default=dynamicdefault,  # to make backporting simpler
 )
 coreconfigitem(
     b'subrepos',
     b'hg:allowed',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'subrepos',
     b'git:allowed',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'subrepos',
     b'svn:allowed',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'templates',
     b'.*',
     default=None,
     generic=True,
 )
 coreconfigitem(
     b'templateconfig',
     b'.*',
     default=dynamicdefault,
     generic=True,
 )
 coreconfigitem(
     b'trusted',
     b'groups',
     default=list,
 )
 coreconfigitem(
     b'trusted',
     b'users',
     default=list,
 )
 coreconfigitem(
     b'ui',
     b'_usedassubrepo',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'allowemptycommit',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'archivemeta',
     default=True,
 )
 coreconfigitem(
     b'ui',
     b'askusername',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'available-memory',
     default=None,
 )
 
 coreconfigitem(
     b'ui',
     b'clonebundlefallback',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'clonebundleprefers',
     default=list,
 )
 coreconfigitem(
     b'ui',
     b'clonebundles',
     default=True,
 )
 coreconfigitem(
     b'ui',
     b'color',
     default=b'auto',
 )
 coreconfigitem(
     b'ui',
     b'commitsubrepos',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'debug',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'debugger',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'editor',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'ui',
     b'detailed-exit-code',
     default=False,
     experimental=True,
 )
 coreconfigitem(
     b'ui',
     b'fallbackencoding',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'forcecwd',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'forcemerge',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'formatdebug',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'formatjson',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'formatted',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'interactive',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'interface',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'interface.chunkselector',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'large-file-limit',
     default=10 * (2 ** 20),
 )
 coreconfigitem(
     b'ui',
     b'logblockedtimes',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'merge',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'mergemarkers',
     default=b'basic',
 )
 coreconfigitem(
     b'ui',
     b'message-output',
     default=b'stdio',
 )
 coreconfigitem(
     b'ui',
     b'nontty',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'origbackuppath',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'paginate',
     default=True,
 )
 coreconfigitem(
     b'ui',
     b'patch',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'portablefilenames',
     default=b'warn',
 )
 coreconfigitem(
     b'ui',
     b'promptecho',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'quiet',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'quietbookmarkmove',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'relative-paths',
     default=b'legacy',
 )
 coreconfigitem(
     b'ui',
     b'remotecmd',
     default=b'hg',
 )
 coreconfigitem(
     b'ui',
     b'report_untrusted',
     default=True,
 )
 coreconfigitem(
     b'ui',
     b'rollback',
     default=True,
 )
 coreconfigitem(
     b'ui',
     b'signal-safe-lock',
     default=True,
 )
 coreconfigitem(
     b'ui',
     b'slash',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'ssh',
     default=b'ssh',
 )
 coreconfigitem(
     b'ui',
     b'ssherrorhint',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'statuscopies',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'strict',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'style',
     default=b'',
 )
 coreconfigitem(
     b'ui',
     b'supportcontact',
     default=None,
 )
 coreconfigitem(
     b'ui',
     b'textwidth',
     default=78,
 )
 coreconfigitem(
     b'ui',
     b'timeout',
     default=b'600',
 )
 coreconfigitem(
     b'ui',
     b'timeout.warn',
     default=0,
 )
 coreconfigitem(
     b'ui',
     b'timestamp-output',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'traceback',
     default=False,
 )
 coreconfigitem(
     b'ui',
     b'tweakdefaults',
     default=False,
 )
 coreconfigitem(b'ui', b'username', alias=[(b'ui', b'user')])
 coreconfigitem(
     b'ui',
     b'verbose',
     default=False,
 )
 coreconfigitem(
     b'verify',
     b'skipflags',
     default=0,
 )
 coreconfigitem(
     b'web',
     b'allowbz2',
     default=False,
 )
 coreconfigitem(
     b'web',
     b'allowgz',
     default=False,
 )
 coreconfigitem(
     b'web',
     b'allow-pull',
     alias=[(b'web', b'allowpull')],
     default=True,
 )
 coreconfigitem(
     b'web',
     b'allow-push',
     alias=[(b'web', b'allow_push')],
     default=list,
 )
 coreconfigitem(
     b'web',
     b'allowzip',
     default=False,
 )
 coreconfigitem(
     b'web',
     b'archivesubrepos',
     default=False,
 )
 coreconfigitem(
     b'web',
     b'cache',
     default=True,
 )
 coreconfigitem(
     b'web',
     b'comparisoncontext',
     default=5,
 )
 coreconfigitem(
     b'web',
     b'contact',
     default=None,
 )
 coreconfigitem(
     b'web',
     b'deny_push',
     default=list,
 )
 coreconfigitem(
     b'web',
     b'guessmime',
     default=False,
 )
 coreconfigitem(
     b'web',
     b'hidden',
     default=False,
 )
 coreconfigitem(
     b'web',
     b'labels',
     default=list,
 )
 coreconfigitem(
     b'web',
     b'logoimg',
     default=b'hglogo.png',
 )
 coreconfigitem(
     b'web',
     b'logourl',
     default=b'https://mercurial-scm.org/',
 )
 coreconfigitem(
     b'web',
     b'accesslog',
     default=b'-',
 )
 coreconfigitem(
     b'web',
     b'address',
     default=b'',
 )
 coreconfigitem(
     b'web',
     b'allow-archive',
     alias=[(b'web', b'allow_archive')],
     default=list,
 )
 coreconfigitem(
     b'web',
     b'allow_read',
     default=list,
 )
 coreconfigitem(
     b'web',
     b'baseurl',
     default=None,
 )
 coreconfigitem(
     b'web',
     b'cacerts',
     default=None,
 )
 coreconfigitem(
     b'web',
     b'certificate',
     default=None,
 )
 coreconfigitem(
     b'web',
     b'collapse',
     default=False,
 )
 coreconfigitem(
     b'web',
     b'csp',
     default=None,
 )
 coreconfigitem(
     b'web',
     b'deny_read',
     default=list,
 )
 coreconfigitem(
     b'web',
     b'descend',
     default=True,
 )
 coreconfigitem(
     b'web',
     b'description',
     default=b"",
 )
 coreconfigitem(
     b'web',
     b'encoding',
     default=lambda: encoding.encoding,
 )
 coreconfigitem(
     b'web',
     b'errorlog',
     default=b'-',
 )
 coreconfigitem(
     b'web',
     b'ipv6',
     default=False,
 )
 coreconfigitem(
     b'web',
     b'maxchanges',
     default=10,
 )
 coreconfigitem(
     b'web',
     b'maxfiles',
     default=10,
 )
 coreconfigitem(
     b'web',
     b'maxshortchanges',
     default=60,
 )
 coreconfigitem(
     b'web',
     b'motd',
     default=b'',
 )
 coreconfigitem(
     b'web',
     b'name',
     default=dynamicdefault,
 )
 coreconfigitem(
     b'web',
     b'port',
     default=8000,
 )
 coreconfigitem(
     b'web',
     b'prefix',
     default=b'',
 )
 coreconfigitem(
     b'web',
     b'push_ssl',
     default=True,
 )
 coreconfigitem(
     b'web',
     b'refreshinterval',
     default=20,
 )
 coreconfigitem(
     b'web',
     b'server-header',
     default=None,
 )
 coreconfigitem(
     b'web',
     b'static',
     default=None,
 )
 coreconfigitem(
     b'web',
     b'staticurl',
     default=None,
 )
 coreconfigitem(
     b'web',
     b'stripes',
     default=1,
 )
 coreconfigitem(
     b'web',
     b'style',
     default=b'paper',
 )
 coreconfigitem(
     b'web',
     b'templates',
     default=None,
 )
 coreconfigitem(
     b'web',
     b'view',
     default=b'served',
     experimental=True,
 )
 coreconfigitem(
     b'worker',
     b'backgroundclose',
     default=dynamicdefault,
 )
 # Windows defaults to a limit of 512 open files. A buffer of 128
 # should give us enough headway.
 coreconfigitem(
     b'worker',
     b'backgroundclosemaxqueue',
     default=384,
 )
 coreconfigitem(
     b'worker',
     b'backgroundcloseminfilecount',
     default=2048,
 )
 coreconfigitem(
     b'worker',
     b'backgroundclosethreadcount',
     default=4,
 )
 coreconfigitem(
     b'worker',
     b'enabled',
     default=True,
 )
 coreconfigitem(
     b'worker',
     b'numcpus',
     default=None,
 )
 
 # Rebase related configuration moved to core because other extension are doing
 # strange things. For example, shelve import the extensions to reuse some bit
 # without formally loading it.
 coreconfigitem(
     b'commands',
     b'rebase.requiredest',
     default=False,
 )
 coreconfigitem(
     b'experimental',
     b'rebaseskipobsolete',
     default=True,
 )
 coreconfigitem(
     b'rebase',
     b'singletransaction',
     default=False,
 )
 coreconfigitem(
     b'rebase',
     b'experimental.inmemory',
     default=False,
 )
 
 # This setting controls creation of a rebase_source extra field
 # during rebase. When False, no such field is created. This is
 # useful eg for incrementally converting changesets and then
 # rebasing them onto an existing repo.
 # WARNING: this is an advanced setting reserved for people who know
 # exactly what they are doing. Misuse of this setting can easily
 # result in obsmarker cycles and a vivid headache.
 coreconfigitem(
     b'rebase',
     b'store-source',
     default=True,
     experimental=True,
 )

mercurial/context.py

@@ -1,3149 +1,3150 b''
 # context.py - changeset and file context objects for mercurial
 #
 # Copyright 2006, 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 filecmp
 import os
 import stat
 
 from .i18n import _
 from .node import (
     hex,
     nullrev,
     short,
 )
 from .pycompat import (
     getattr,
 )
 from . import (
     dagop,
     encoding,
     error,
     fileset,
     match as matchmod,
     mergestate as mergestatemod,
     metadata,
     obsolete as obsmod,
     patch,
     pathutil,
     phases,
     repoview,
     scmutil,
     sparse,
     subrepo,
     subrepoutil,
+    testing,
     util,
 )
 from .utils import (
     dateutil,
     stringutil,
 )
 from .dirstateutils import (
     timestamp,
 )
 
 propertycache = util.propertycache
 
 
 class basectx:
     """A basectx object represents the common logic for its children:
     changectx: read-only context that is already present in the repo,
     workingctx: a context that represents the working directory and can
                 be committed,
     memctx: a context that represents changes in-memory and can also
             be committed."""
 
     def __init__(self, repo):
         self._repo = repo
 
     def __bytes__(self):
         return short(self.node())
 
     __str__ = encoding.strmethod(__bytes__)
 
     def __repr__(self):
         return "<%s %s>" % (type(self).__name__, str(self))
 
     def __eq__(self, other):
         try:
             return type(self) == type(other) and self._rev == other._rev
         except AttributeError:
             return False
 
     def __ne__(self, other):
         return not (self == other)
 
     def __contains__(self, key):
         return key in self._manifest
 
     def __getitem__(self, key):
         return self.filectx(key)
 
     def __iter__(self):
         return iter(self._manifest)
 
     def _buildstatusmanifest(self, status):
         """Builds a manifest that includes the given status results, if this is
         a working copy context. For non-working copy contexts, it just returns
         the normal manifest."""
         return self.manifest()
 
     def _matchstatus(self, other, match):
         """This internal method provides a way for child objects to override the
         match operator.
         """
         return match
 
     def _buildstatus(
         self, other, s, match, listignored, listclean, listunknown
     ):
         """build a status with respect to another context"""
         # Load earliest manifest first for caching reasons. More specifically,
         # if you have revisions 1000 and 1001, 1001 is probably stored as a
         # delta against 1000. Thus, if you read 1000 first, we'll reconstruct
         # 1000 and cache it so that when you read 1001, we just need to apply a
         # delta to what's in the cache. So that's one full reconstruction + one
         # delta application.
         mf2 = None
         if self.rev() is not None and self.rev() < other.rev():
             mf2 = self._buildstatusmanifest(s)
         mf1 = other._buildstatusmanifest(s)
         if mf2 is None:
             mf2 = self._buildstatusmanifest(s)
 
         modified, added = [], []
         removed = []
         clean = []
         deleted, unknown, ignored = s.deleted, s.unknown, s.ignored
         deletedset = set(deleted)
         d = mf1.diff(mf2, match=match, clean=listclean)
         for fn, value in d.items():
             if fn in deletedset:
                 continue
             if value is None:
                 clean.append(fn)
                 continue
             (node1, flag1), (node2, flag2) = value
             if node1 is None:
                 added.append(fn)
             elif node2 is None:
                 removed.append(fn)
             elif flag1 != flag2:
                 modified.append(fn)
             elif node2 not in self._repo.nodeconstants.wdirfilenodeids:
                 # When comparing files between two commits, we save time by
                 # not comparing the file contents when the nodeids differ.
                 # Note that this means we incorrectly report a reverted change
                 # to a file as a modification.
                 modified.append(fn)
             elif self[fn].cmp(other[fn]):
                 modified.append(fn)
             else:
                 clean.append(fn)
 
         if removed:
             # need to filter files if they are already reported as removed
             unknown = [
                 fn
                 for fn in unknown
                 if fn not in mf1 and (not match or match(fn))
             ]
             ignored = [
                 fn
                 for fn in ignored
                 if fn not in mf1 and (not match or match(fn))
             ]
             # if they're deleted, don't report them as removed
             removed = [fn for fn in removed if fn not in deletedset]
 
         return scmutil.status(
             modified, added, removed, deleted, unknown, ignored, clean
         )
 
     @propertycache
     def substate(self):
         return subrepoutil.state(self, self._repo.ui)
 
     def subrev(self, subpath):
         return self.substate[subpath][1]
 
     def rev(self):
         return self._rev
 
     def node(self):
         return self._node
 
     def hex(self):
         return hex(self.node())
 
     def manifest(self):
         return self._manifest
 
     def manifestctx(self):
         return self._manifestctx
 
     def repo(self):
         return self._repo
 
     def phasestr(self):
         return phases.phasenames[self.phase()]
 
     def mutable(self):
         return self.phase() > phases.public
 
     def matchfileset(self, cwd, expr, badfn=None):
         return fileset.match(self, cwd, expr, badfn=badfn)
 
     def obsolete(self):
         """True if the changeset is obsolete"""
         return self.rev() in obsmod.getrevs(self._repo, b'obsolete')
 
     def extinct(self):
         """True if the changeset is extinct"""
         return self.rev() in obsmod.getrevs(self._repo, b'extinct')
 
     def orphan(self):
         """True if the changeset is not obsolete, but its ancestor is"""
         return self.rev() in obsmod.getrevs(self._repo, b'orphan')
 
     def phasedivergent(self):
         """True if the changeset tries to be a successor of a public changeset
 
         Only non-public and non-obsolete changesets may be phase-divergent.
         """
         return self.rev() in obsmod.getrevs(self._repo, b'phasedivergent')
 
     def contentdivergent(self):
         """Is a successor of a changeset with multiple possible successor sets
 
         Only non-public and non-obsolete changesets may be content-divergent.
         """
         return self.rev() in obsmod.getrevs(self._repo, b'contentdivergent')
 
     def isunstable(self):
         """True if the changeset is either orphan, phase-divergent or
         content-divergent"""
         return self.orphan() or self.phasedivergent() or self.contentdivergent()
 
     def instabilities(self):
         """return the list of instabilities affecting this changeset.
 
         Instabilities are returned as strings. possible values are:
         - orphan,
         - phase-divergent,
         - content-divergent.
         """
         instabilities = []
         if self.orphan():
             instabilities.append(b'orphan')
         if self.phasedivergent():
             instabilities.append(b'phase-divergent')
         if self.contentdivergent():
             instabilities.append(b'content-divergent')
         return instabilities
 
     def parents(self):
         """return contexts for each parent changeset"""
         return self._parents
 
     def p1(self):
         return self._parents[0]
 
     def p2(self):
         parents = self._parents
         if len(parents) == 2:
             return parents[1]
         return self._repo[nullrev]
 
     def _fileinfo(self, path):
         if '_manifest' in self.__dict__:
             try:
                 return self._manifest.find(path)
             except KeyError:
                 raise error.ManifestLookupError(
                     self._node or b'None', path, _(b'not found in manifest')
                 )
         if '_manifestdelta' in self.__dict__ or path in self.files():
             if path in self._manifestdelta:
                 return (
                     self._manifestdelta[path],
                     self._manifestdelta.flags(path),
                 )
         mfl = self._repo.manifestlog
         try:
             node, flag = mfl[self._changeset.manifest].find(path)
         except KeyError:
             raise error.ManifestLookupError(
                 self._node or b'None', path, _(b'not found in manifest')
             )
 
         return node, flag
 
     def filenode(self, path):
         return self._fileinfo(path)[0]
 
     def flags(self, path):
         try:
             return self._fileinfo(path)[1]
         except error.LookupError:
             return b''
 
     @propertycache
     def _copies(self):
         return metadata.computechangesetcopies(self)
 
     def p1copies(self):
         return self._copies[0]
 
     def p2copies(self):
         return self._copies[1]
 
     def sub(self, path, allowcreate=True):
         '''return a subrepo for the stored revision of path, never wdir()'''
         return subrepo.subrepo(self, path, allowcreate=allowcreate)
 
     def nullsub(self, path, pctx):
         return subrepo.nullsubrepo(self, path, pctx)
 
     def workingsub(self, path):
         """return a subrepo for the stored revision, or wdir if this is a wdir
         context.
         """
         return subrepo.subrepo(self, path, allowwdir=True)
 
     def match(
         self,
         pats=None,
         include=None,
         exclude=None,
         default=b'glob',
         listsubrepos=False,
         badfn=None,
         cwd=None,
     ):
         r = self._repo
         if not cwd:
             cwd = r.getcwd()
         return matchmod.match(
             r.root,
             cwd,
             pats,
             include,
             exclude,
             default,
             auditor=r.nofsauditor,
             ctx=self,
             listsubrepos=listsubrepos,
             badfn=badfn,
         )
 
     def diff(
         self,
         ctx2=None,
         match=None,
         changes=None,
         opts=None,
         losedatafn=None,
         pathfn=None,
         copy=None,
         copysourcematch=None,
         hunksfilterfn=None,
     ):
         """Returns a diff generator for the given contexts and matcher"""
         if ctx2 is None:
             ctx2 = self.p1()
         if ctx2 is not None:
             ctx2 = self._repo[ctx2]
         return patch.diff(
             self._repo,
             ctx2,
             self,
             match=match,
             changes=changes,
             opts=opts,
             losedatafn=losedatafn,
             pathfn=pathfn,
             copy=copy,
             copysourcematch=copysourcematch,
             hunksfilterfn=hunksfilterfn,
         )
 
     def dirs(self):
         return self._manifest.dirs()
 
     def hasdir(self, dir):
         return self._manifest.hasdir(dir)
 
     def status(
         self,
         other=None,
         match=None,
         listignored=False,
         listclean=False,
         listunknown=False,
         listsubrepos=False,
     ):
         """return status of files between two nodes or node and working
         directory.
 
         If other is None, compare this node with working directory.
 
         ctx1.status(ctx2) returns the status of change from ctx1 to ctx2
 
         Returns a mercurial.scmutils.status object.
 
         Data can be accessed using either tuple notation:
 
             (modified, added, removed, deleted, unknown, ignored, clean)
 
         or direct attribute access:
 
             s.modified, s.added, ...
         """
 
         ctx1 = self
         ctx2 = self._repo[other]
 
         # This next code block is, admittedly, fragile logic that tests for
         # reversing the contexts and wouldn't need to exist if it weren't for
         # the fast (and common) code path of comparing the working directory
         # with its first parent.
         #
         # What we're aiming for here is the ability to call:
         #
         # workingctx.status(parentctx)
         #
         # If we always built the manifest for each context and compared those,
         # then we'd be done. But the special case of the above call means we
         # just copy the manifest of the parent.
         reversed = False
         if not isinstance(ctx1, changectx) and isinstance(ctx2, changectx):
             reversed = True
             ctx1, ctx2 = ctx2, ctx1
 
         match = self._repo.narrowmatch(match)
         match = ctx2._matchstatus(ctx1, match)
         r = scmutil.status([], [], [], [], [], [], [])
         r = ctx2._buildstatus(
             ctx1, r, match, listignored, listclean, listunknown
         )
 
         if reversed:
             # Reverse added and removed. Clear deleted, unknown and ignored as
             # these make no sense to reverse.
             r = scmutil.status(
                 r.modified, r.removed, r.added, [], [], [], r.clean
             )
 
         if listsubrepos:
             for subpath, sub in scmutil.itersubrepos(ctx1, ctx2):
                 try:
                     rev2 = ctx2.subrev(subpath)
                 except KeyError:
                     # A subrepo that existed in node1 was deleted between
                     # node1 and node2 (inclusive). Thus, ctx2's substate
                     # won't contain that subpath. The best we can do ignore it.
                     rev2 = None
                 submatch = matchmod.subdirmatcher(subpath, match)
                 s = sub.status(
                     rev2,
                     match=submatch,
                     ignored=listignored,
                     clean=listclean,
                     unknown=listunknown,
                     listsubrepos=True,
                 )
                 for k in (
                     'modified',
                     'added',
                     'removed',
                     'deleted',
                     'unknown',
                     'ignored',
                     'clean',
                 ):
                     rfiles, sfiles = getattr(r, k), getattr(s, k)
                     rfiles.extend(b"%s/%s" % (subpath, f) for f in sfiles)
 
         r.modified.sort()
         r.added.sort()
         r.removed.sort()
         r.deleted.sort()
         r.unknown.sort()
         r.ignored.sort()
         r.clean.sort()
 
         return r
 
     def mergestate(self, clean=False):
         """Get a mergestate object for this context."""
         raise NotImplementedError(
             '%s does not implement mergestate()' % self.__class__
         )
 
     def isempty(self):
         return not (
             len(self.parents()) > 1
             or self.branch() != self.p1().branch()
             or self.closesbranch()
             or self.files()
         )
 
 
 class changectx(basectx):
     """A changecontext object makes access to data related to a particular
     changeset convenient. It represents a read-only context already present in
     the repo."""
 
     def __init__(self, repo, rev, node, maybe_filtered=True):
         super(changectx, self).__init__(repo)
         self._rev = rev
         self._node = node
         # When maybe_filtered is True, the revision might be affected by
         # changelog filtering and operation through the filtered changelog must be used.
         #
         # When maybe_filtered is False, the revision has already been checked
         # against filtering and is not filtered. Operation through the
         # unfiltered changelog might be used in some case.
         self._maybe_filtered = maybe_filtered
 
     def __hash__(self):
         try:
             return hash(self._rev)
         except AttributeError:
             return id(self)
 
     def __nonzero__(self):
         return self._rev != nullrev
 
     __bool__ = __nonzero__
 
     @propertycache
     def _changeset(self):
         if self._maybe_filtered:
             repo = self._repo
         else:
             repo = self._repo.unfiltered()
         return repo.changelog.changelogrevision(self.rev())
 
     @propertycache
     def _manifest(self):
         return self._manifestctx.read()
 
     @property
     def _manifestctx(self):
         return self._repo.manifestlog[self._changeset.manifest]
 
     @propertycache
     def _manifestdelta(self):
         return self._manifestctx.readdelta()
 
     @propertycache
     def _parents(self):
         repo = self._repo
         if self._maybe_filtered:
             cl = repo.changelog
         else:
             cl = repo.unfiltered().changelog
 
         p1, p2 = cl.parentrevs(self._rev)
         if p2 == nullrev:
             return [changectx(repo, p1, cl.node(p1), maybe_filtered=False)]
         return [
             changectx(repo, p1, cl.node(p1), maybe_filtered=False),
             changectx(repo, p2, cl.node(p2), maybe_filtered=False),
         ]
 
     def changeset(self):
         c = self._changeset
         return (
             c.manifest,
             c.user,
             c.date,
             c.files,
             c.description,
             c.extra,
         )
 
     def manifestnode(self):
         return self._changeset.manifest
 
     def user(self):
         return self._changeset.user
 
     def date(self):
         return self._changeset.date
 
     def files(self):
         return self._changeset.files
 
     def filesmodified(self):
         modified = set(self.files())
         modified.difference_update(self.filesadded())
         modified.difference_update(self.filesremoved())
         return sorted(modified)
 
     def filesadded(self):
         filesadded = self._changeset.filesadded
         compute_on_none = True
         if self._repo.filecopiesmode == b'changeset-sidedata':
             compute_on_none = False
         else:
             source = self._repo.ui.config(b'experimental', b'copies.read-from')
             if source == b'changeset-only':
                 compute_on_none = False
             elif source != b'compatibility':
                 # filelog mode, ignore any changelog content
                 filesadded = None
         if filesadded is None:
             if compute_on_none:
                 filesadded = metadata.computechangesetfilesadded(self)
             else:
                 filesadded = []
         return filesadded
 
     def filesremoved(self):
         filesremoved = self._changeset.filesremoved
         compute_on_none = True
         if self._repo.filecopiesmode == b'changeset-sidedata':
             compute_on_none = False
         else:
             source = self._repo.ui.config(b'experimental', b'copies.read-from')
             if source == b'changeset-only':
                 compute_on_none = False
             elif source != b'compatibility':
                 # filelog mode, ignore any changelog content
                 filesremoved = None
         if filesremoved is None:
             if compute_on_none:
                 filesremoved = metadata.computechangesetfilesremoved(self)
             else:
                 filesremoved = []
         return filesremoved
 
     @propertycache
     def _copies(self):
         p1copies = self._changeset.p1copies
         p2copies = self._changeset.p2copies
         compute_on_none = True
         if self._repo.filecopiesmode == b'changeset-sidedata':
             compute_on_none = False
         else:
             source = self._repo.ui.config(b'experimental', b'copies.read-from')
             # If config says to get copy metadata only from changeset, then
             # return that, defaulting to {} if there was no copy metadata.  In
             # compatibility mode, we return copy data from the changeset if it
             # was recorded there, and otherwise we fall back to getting it from
             # the filelogs (below).
             #
             # If we are in compatiblity mode and there is not data in the
             # changeset), we get the copy metadata from the filelogs.
             #
             # otherwise, when config said to read only from filelog, we get the
             # copy metadata from the filelogs.
             if source == b'changeset-only':
                 compute_on_none = False
             elif source != b'compatibility':
                 # filelog mode, ignore any changelog content
                 p1copies = p2copies = None
         if p1copies is None:
             if compute_on_none:
                 p1copies, p2copies = super(changectx, self)._copies
             else:
                 if p1copies is None:
                     p1copies = {}
         if p2copies is None:
             p2copies = {}
         return p1copies, p2copies
 
     def description(self):
         return self._changeset.description
 
     def branch(self):
         return encoding.tolocal(self._changeset.extra.get(b"branch"))
 
     def closesbranch(self):
         return b'close' in self._changeset.extra
 
     def extra(self):
         """Return a dict of extra information."""
         return self._changeset.extra
 
     def tags(self):
         """Return a list of byte tag names"""
         return self._repo.nodetags(self._node)
 
     def bookmarks(self):
         """Return a list of byte bookmark names."""
         return self._repo.nodebookmarks(self._node)
 
     def fast_rank(self):
         repo = self._repo
         if self._maybe_filtered:
             cl = repo.changelog
         else:
             cl = repo.unfiltered().changelog
         return cl.fast_rank(self._rev)
 
     def phase(self):
         return self._repo._phasecache.phase(self._repo, self._rev)
 
     def hidden(self):
         return self._rev in repoview.filterrevs(self._repo, b'visible')
 
     def isinmemory(self):
         return False
 
     def children(self):
         """return list of changectx contexts for each child changeset.
 
         This returns only the immediate child changesets. Use descendants() to
         recursively walk children.
         """
         c = self._repo.changelog.children(self._node)
         return [self._repo[x] for x in c]
 
     def ancestors(self):
         for a in self._repo.changelog.ancestors([self._rev]):
             yield self._repo[a]
 
     def descendants(self):
         """Recursively yield all children of the changeset.
 
         For just the immediate children, use children()
         """
         for d in self._repo.changelog.descendants([self._rev]):
             yield self._repo[d]
 
     def filectx(self, path, fileid=None, filelog=None):
         """get a file context from this changeset"""
         if fileid is None:
             fileid = self.filenode(path)
         return filectx(
             self._repo, path, fileid=fileid, changectx=self, filelog=filelog
         )
 
     def ancestor(self, c2, warn=False):
         """return the "best" ancestor context of self and c2
 
         If there are multiple candidates, it will show a message and check
         merge.preferancestor configuration before falling back to the
         revlog ancestor."""
         # deal with workingctxs
         n2 = c2._node
         if n2 is None:
             n2 = c2._parents[0]._node
         cahs = self._repo.changelog.commonancestorsheads(self._node, n2)
         if not cahs:
             anc = self._repo.nodeconstants.nullid
         elif len(cahs) == 1:
             anc = cahs[0]
         else:
             # experimental config: merge.preferancestor
             for r in self._repo.ui.configlist(b'merge', b'preferancestor'):
                 try:
                     ctx = scmutil.revsymbol(self._repo, r)
                 except error.RepoLookupError:
                     continue
                 anc = ctx.node()
                 if anc in cahs:
                     break
             else:
                 anc = self._repo.changelog.ancestor(self._node, n2)
             if warn:
                 self._repo.ui.status(
                     (
                         _(b"note: using %s as ancestor of %s and %s\n")
                         % (short(anc), short(self._node), short(n2))
                     )
                     + b''.join(
                         _(
                             b"      alternatively, use --config "
                             b"merge.preferancestor=%s\n"
                         )
                         % short(n)
                         for n in sorted(cahs)
                         if n != anc
                     )
                 )
         return self._repo[anc]
 
     def isancestorof(self, other):
         """True if this changeset is an ancestor of other"""
         return self._repo.changelog.isancestorrev(self._rev, other._rev)
 
     def walk(self, match):
         '''Generates matching file names.'''
 
         # Wrap match.bad method to have message with nodeid
         def bad(fn, msg):
             # The manifest doesn't know about subrepos, so don't complain about
             # paths into valid subrepos.
             if any(fn == s or fn.startswith(s + b'/') for s in self.substate):
                 return
             match.bad(fn, _(b'no such file in rev %s') % self)
 
         m = matchmod.badmatch(self._repo.narrowmatch(match), bad)
         return self._manifest.walk(m)
 
     def matches(self, match):
         return self.walk(match)
 
 
 class basefilectx:
     """A filecontext object represents the common logic for its children:
     filectx: read-only access to a filerevision that is already present
              in the repo,
     workingfilectx: a filecontext that represents files from the working
                     directory,
     memfilectx: a filecontext that represents files in-memory,
     """
 
     @propertycache
     def _filelog(self):
         return self._repo.file(self._path)
 
     @propertycache
     def _changeid(self):
         if '_changectx' in self.__dict__:
             return self._changectx.rev()
         elif '_descendantrev' in self.__dict__:
             # this file context was created from a revision with a known
             # descendant, we can (lazily) correct for linkrev aliases
             return self._adjustlinkrev(self._descendantrev)
         else:
             return self._filelog.linkrev(self._filerev)
 
     @propertycache
     def _filenode(self):
         if '_fileid' in self.__dict__:
             return self._filelog.lookup(self._fileid)
         else:
             return self._changectx.filenode(self._path)
 
     @propertycache
     def _filerev(self):
         return self._filelog.rev(self._filenode)
 
     @propertycache
     def _repopath(self):
         return self._path
 
     def __nonzero__(self):
         try:
             self._filenode
             return True
         except error.LookupError:
             # file is missing
             return False
 
     __bool__ = __nonzero__
 
     def __bytes__(self):
         try:
             return b"%s@%s" % (self.path(), self._changectx)
         except error.LookupError:
             return b"%s@???" % self.path()
 
     __str__ = encoding.strmethod(__bytes__)
 
     def __repr__(self):
         return "<%s %s>" % (type(self).__name__, str(self))
 
     def __hash__(self):
         try:
             return hash((self._path, self._filenode))
         except AttributeError:
             return id(self)
 
     def __eq__(self, other):
         try:
             return (
                 type(self) == type(other)
                 and self._path == other._path
                 and self._filenode == other._filenode
             )
         except AttributeError:
             return False
 
     def __ne__(self, other):
         return not (self == other)
 
     def filerev(self):
         return self._filerev
 
     def filenode(self):
         return self._filenode
 
     @propertycache
     def _flags(self):
         return self._changectx.flags(self._path)
 
     def flags(self):
         return self._flags
 
     def filelog(self):
         return self._filelog
 
     def rev(self):
         return self._changeid
 
     def linkrev(self):
         return self._filelog.linkrev(self._filerev)
 
     def node(self):
         return self._changectx.node()
 
     def hex(self):
         return self._changectx.hex()
 
     def user(self):
         return self._changectx.user()
 
     def date(self):
         return self._changectx.date()
 
     def files(self):
         return self._changectx.files()
 
     def description(self):
         return self._changectx.description()
 
     def branch(self):
         return self._changectx.branch()
 
     def extra(self):
         return self._changectx.extra()
 
     def phase(self):
         return self._changectx.phase()
 
     def phasestr(self):
         return self._changectx.phasestr()
 
     def obsolete(self):
         return self._changectx.obsolete()
 
     def instabilities(self):
         return self._changectx.instabilities()
 
     def manifest(self):
         return self._changectx.manifest()
 
     def changectx(self):
         return self._changectx
 
     def renamed(self):
         return self._copied
 
     def copysource(self):
         return self._copied and self._copied[0]
 
     def repo(self):
         return self._repo
 
     def size(self):
         return len(self.data())
 
     def path(self):
         return self._path
 
     def isbinary(self):
         try:
             return stringutil.binary(self.data())
         except IOError:
             return False
 
     def isexec(self):
         return b'x' in self.flags()
 
     def islink(self):
         return b'l' in self.flags()
 
     def isabsent(self):
         """whether this filectx represents a file not in self._changectx
 
         This is mainly for merge code to detect change/delete conflicts. This is
         expected to be True for all subclasses of basectx."""
         return False
 
     _customcmp = False
 
     def cmp(self, fctx):
         """compare with other file context
 
         returns True if different than fctx.
         """
         if fctx._customcmp:
             return fctx.cmp(self)
 
         if self._filenode is None:
             raise error.ProgrammingError(
                 b'filectx.cmp() must be reimplemented if not backed by revlog'
             )
 
         if fctx._filenode is None:
             if self._repo._encodefilterpats:
                 # can't rely on size() because wdir content may be decoded
                 return self._filelog.cmp(self._filenode, fctx.data())
             # filelog.size() has two special cases:
             # - censored metadata
             # - copy/rename tracking
             # The first is detected by peaking into the delta,
             # the second is detected by abusing parent order
             # in the revlog index as flag bit. This leaves files using
             # the dummy encoding and non-standard meta attributes.
             # The following check is a special case for the empty
             # metadata block used if the raw file content starts with '\1\n'.
             # Cases of arbitrary metadata flags are currently mishandled.
             if self.size() - 4 == fctx.size():
                 # size() can match:
                 # if file data starts with '\1\n', empty metadata block is
                 # prepended, which adds 4 bytes to filelog.size().
                 return self._filelog.cmp(self._filenode, fctx.data())
         if self.size() == fctx.size() or self.flags() == b'l':
             # size() matches: need to compare content
             # issue6456: Always compare symlinks because size can represent
             # encrypted string for EXT-4 encryption(fscrypt).
             return self._filelog.cmp(self._filenode, fctx.data())
 
         # size() differs
         return True
 
     def _adjustlinkrev(self, srcrev, inclusive=False, stoprev=None):
         """return the first ancestor of <srcrev> introducing <fnode>
 
         If the linkrev of the file revision does not point to an ancestor of
         srcrev, we'll walk down the ancestors until we find one introducing
         this file revision.
 
         :srcrev: the changeset revision we search ancestors from
         :inclusive: if true, the src revision will also be checked
         :stoprev: an optional revision to stop the walk at. If no introduction
                   of this file content could be found before this floor
                   revision, the function will returns "None" and stops its
                   iteration.
         """
         repo = self._repo
         cl = repo.unfiltered().changelog
         mfl = repo.manifestlog
         # fetch the linkrev
         lkr = self.linkrev()
         if srcrev == lkr:
             return lkr
         # hack to reuse ancestor computation when searching for renames
         memberanc = getattr(self, '_ancestrycontext', None)
         iteranc = None
         if srcrev is None:
             # wctx case, used by workingfilectx during mergecopy
             revs = [p.rev() for p in self._repo[None].parents()]
             inclusive = True  # we skipped the real (revless) source
         else:
             revs = [srcrev]
         if memberanc is None:
             memberanc = iteranc = cl.ancestors(revs, lkr, inclusive=inclusive)
         # check if this linkrev is an ancestor of srcrev
         if lkr not in memberanc:
             if iteranc is None:
                 iteranc = cl.ancestors(revs, lkr, inclusive=inclusive)
             fnode = self._filenode
             path = self._path
             for a in iteranc:
                 if stoprev is not None and a < stoprev:
                     return None
                 ac = cl.read(a)  # get changeset data (we avoid object creation)
                 if path in ac[3]:  # checking the 'files' field.
                     # The file has been touched, check if the content is
                     # similar to the one we search for.
                     if fnode == mfl[ac[0]].readfast().get(path):
                         return a
             # In theory, we should never get out of that loop without a result.
             # But if manifest uses a buggy file revision (not children of the
             # one it replaces) we could. Such a buggy situation will likely
             # result is crash somewhere else at to some point.
         return lkr
 
     def isintroducedafter(self, changelogrev):
         """True if a filectx has been introduced after a given floor revision"""
         if self.linkrev() >= changelogrev:
             return True
         introrev = self._introrev(stoprev=changelogrev)
         if introrev is None:
             return False
         return introrev >= changelogrev
 
     def introrev(self):
         """return the rev of the changeset which introduced this file revision
 
         This method is different from linkrev because it take into account the
         changeset the filectx was created from. It ensures the returned
         revision is one of its ancestors. This prevents bugs from
         'linkrev-shadowing' when a file revision is used by multiple
         changesets.
         """
         return self._introrev()
 
     def _introrev(self, stoprev=None):
         """
         Same as `introrev` but, with an extra argument to limit changelog
         iteration range in some internal usecase.
 
         If `stoprev` is set, the `introrev` will not be searched past that
         `stoprev` revision and "None" might be returned. This is useful to
         limit the iteration range.
         """
         toprev = None
         attrs = vars(self)
         if '_changeid' in attrs:
             # We have a cached value already
             toprev = self._changeid
         elif '_changectx' in attrs:
             # We know which changelog entry we are coming from
             toprev = self._changectx.rev()
 
         if toprev is not None:
             return self._adjustlinkrev(toprev, inclusive=True, stoprev=stoprev)
         elif '_descendantrev' in attrs:
             introrev = self._adjustlinkrev(self._descendantrev, stoprev=stoprev)
             # be nice and cache the result of the computation
             if introrev is not None:
                 self._changeid = introrev
             return introrev
         else:
             return self.linkrev()
 
     def introfilectx(self):
         """Return filectx having identical contents, but pointing to the
         changeset revision where this filectx was introduced"""
         introrev = self.introrev()
         if self.rev() == introrev:
             return self
         return self.filectx(self.filenode(), changeid=introrev)
 
     def _parentfilectx(self, path, fileid, filelog):
         """create parent filectx keeping ancestry info for _adjustlinkrev()"""
         fctx = filectx(self._repo, path, fileid=fileid, filelog=filelog)
         if '_changeid' in vars(self) or '_changectx' in vars(self):
             # If self is associated with a changeset (probably explicitly
             # fed), ensure the created filectx is associated with a
             # changeset that is an ancestor of self.changectx.
             # This lets us later use _adjustlinkrev to get a correct link.
             fctx._descendantrev = self.rev()
             fctx._ancestrycontext = getattr(self, '_ancestrycontext', None)
         elif '_descendantrev' in vars(self):
             # Otherwise propagate _descendantrev if we have one associated.
             fctx._descendantrev = self._descendantrev
             fctx._ancestrycontext = getattr(self, '_ancestrycontext', None)
         return fctx
 
     def parents(self):
         _path = self._path
         fl = self._filelog
         parents = self._filelog.parents(self._filenode)
         pl = [
             (_path, node, fl)
             for node in parents
             if node != self._repo.nodeconstants.nullid
         ]
 
         r = fl.renamed(self._filenode)
         if r:
             # - In the simple rename case, both parent are nullid, pl is empty.
             # - In case of merge, only one of the parent is null id and should
             # be replaced with the rename information. This parent is -always-
             # the first one.
             #
             # As null id have always been filtered out in the previous list
             # comprehension, inserting to 0 will always result in "replacing
             # first nullid parent with rename information.
             pl.insert(0, (r[0], r[1], self._repo.file(r[0])))
 
         return [self._parentfilectx(path, fnode, l) for path, fnode, l in pl]
 
     def p1(self):
         return self.parents()[0]
 
     def p2(self):
         p = self.parents()
         if len(p) == 2:
             return p[1]
         return filectx(self._repo, self._path, fileid=-1, filelog=self._filelog)
 
     def annotate(self, follow=False, skiprevs=None, diffopts=None):
         """Returns a list of annotateline objects for each line in the file
 
         - line.fctx is the filectx of the node where that line was last changed
         - line.lineno is the line number at the first appearance in the managed
           file
         - line.text is the data on that line (including newline character)
         """
         getlog = util.lrucachefunc(lambda x: self._repo.file(x))
 
         def parents(f):
             # Cut _descendantrev here to mitigate the penalty of lazy linkrev
             # adjustment. Otherwise, p._adjustlinkrev() would walk changelog
             # from the topmost introrev (= srcrev) down to p.linkrev() if it
             # isn't an ancestor of the srcrev.
             f._changeid
             pl = f.parents()
 
             # Don't return renamed parents if we aren't following.
             if not follow:
                 pl = [p for p in pl if p.path() == f.path()]
 
             # renamed filectx won't have a filelog yet, so set it
             # from the cache to save time
             for p in pl:
                 if not '_filelog' in p.__dict__:
                     p._filelog = getlog(p.path())
 
             return pl
 
         # use linkrev to find the first changeset where self appeared
         base = self.introfilectx()
         if getattr(base, '_ancestrycontext', None) is None:
             # it is safe to use an unfiltered repository here because we are
             # walking ancestors only.
             cl = self._repo.unfiltered().changelog
             if base.rev() is None:
                 # wctx is not inclusive, but works because _ancestrycontext
                 # is used to test filelog revisions
                 ac = cl.ancestors(
                     [p.rev() for p in base.parents()], inclusive=True
                 )
             else:
                 ac = cl.ancestors([base.rev()], inclusive=True)
             base._ancestrycontext = ac
 
         return dagop.annotate(
             base, parents, skiprevs=skiprevs, diffopts=diffopts
         )
 
     def ancestors(self, followfirst=False):
         visit = {}
         c = self
         if followfirst:
             cut = 1
         else:
             cut = None
 
         while True:
             for parent in c.parents()[:cut]:
                 visit[(parent.linkrev(), parent.filenode())] = parent
             if not visit:
                 break
             c = visit.pop(max(visit))
             yield c
 
     def decodeddata(self):
         """Returns `data()` after running repository decoding filters.
 
         This is often equivalent to how the data would be expressed on disk.
         """
         return self._repo.wwritedata(self.path(), self.data())
 
 
 class filectx(basefilectx):
     """A filecontext object makes access to data related to a particular
     filerevision convenient."""
 
     def __init__(
         self,
         repo,
         path,
         changeid=None,
         fileid=None,
         filelog=None,
         changectx=None,
     ):
         """changeid must be a revision number, if specified.
         fileid can be a file revision or node."""
         self._repo = repo
         self._path = path
 
         assert (
             changeid is not None or fileid is not None or changectx is not None
         ), b"bad args: changeid=%r, fileid=%r, changectx=%r" % (
             changeid,
             fileid,
             changectx,
         )
 
         if filelog is not None:
             self._filelog = filelog
 
         if changeid is not None:
             self._changeid = changeid
         if changectx is not None:
             self._changectx = changectx
         if fileid is not None:
             self._fileid = fileid
 
     @propertycache
     def _changectx(self):
         try:
             return self._repo[self._changeid]
         except error.FilteredRepoLookupError:
             # Linkrev may point to any revision in the repository.  When the
             # repository is filtered this may lead to `filectx` trying to build
             # `changectx` for filtered revision. In such case we fallback to
             # creating `changectx` on the unfiltered version of the reposition.
             # This fallback should not be an issue because `changectx` from
             # `filectx` are not used in complex operations that care about
             # filtering.
             #
             # This fallback is a cheap and dirty fix that prevent several
             # crashes. It does not ensure the behavior is correct. However the
             # behavior was not correct before filtering either and "incorrect
             # behavior" is seen as better as "crash"
             #
             # Linkrevs have several serious troubles with filtering that are
             # complicated to solve. Proper handling of the issue here should be
             # considered when solving linkrev issue are on the table.
             return self._repo.unfiltered()[self._changeid]
 
     def filectx(self, fileid, changeid=None):
         """opens an arbitrary revision of the file without
         opening a new filelog"""
         return filectx(
             self._repo,
             self._path,
             fileid=fileid,
             filelog=self._filelog,
             changeid=changeid,
         )
 
     def rawdata(self):
         return self._filelog.rawdata(self._filenode)
 
     def rawflags(self):
         """low-level revlog flags"""
         return self._filelog.flags(self._filerev)
 
     def data(self):
         try:
             return self._filelog.read(self._filenode)
         except error.CensoredNodeError:
             if self._repo.ui.config(b"censor", b"policy") == b"ignore":
                 return b""
             raise error.Abort(
                 _(b"censored node: %s") % short(self._filenode),
                 hint=_(b"set censor.policy to ignore errors"),
             )
 
     def size(self):
         return self._filelog.size(self._filerev)
 
     @propertycache
     def _copied(self):
         """check if file was actually renamed in this changeset revision
 
         If rename logged in file revision, we report copy for changeset only
         if file revisions linkrev points back to the changeset in question
         or both changeset parents contain different file revisions.
         """
 
         renamed = self._filelog.renamed(self._filenode)
         if not renamed:
             return None
 
         if self.rev() == self.linkrev():
             return renamed
 
         name = self.path()
         fnode = self._filenode
         for p in self._changectx.parents():
             try:
                 if fnode == p.filenode(name):
                     return None
             except error.LookupError:
                 pass
         return renamed
 
     def children(self):
         # hard for renames
         c = self._filelog.children(self._filenode)
         return [
             filectx(self._repo, self._path, fileid=x, filelog=self._filelog)
             for x in c
         ]
 
 
 class committablectx(basectx):
     """A committablectx object provides common functionality for a context that
     wants the ability to commit, e.g. workingctx or memctx."""
 
     def __init__(
         self,
         repo,
         text=b"",
         user=None,
         date=None,
         extra=None,
         changes=None,
         branch=None,
     ):
         super(committablectx, self).__init__(repo)
         self._rev = None
         self._node = None
         self._text = text
         if date:
             self._date = dateutil.parsedate(date)
         if user:
             self._user = user
         if changes:
             self._status = changes
 
         self._extra = {}
         if extra:
             self._extra = extra.copy()
         if branch is not None:
             self._extra[b'branch'] = encoding.fromlocal(branch)
         if not self._extra.get(b'branch'):
             self._extra[b'branch'] = b'default'
 
     def __bytes__(self):
         return bytes(self._parents[0]) + b"+"
 
     def hex(self):
         self._repo.nodeconstants.wdirhex
 
     __str__ = encoding.strmethod(__bytes__)
 
     def __nonzero__(self):
         return True
 
     __bool__ = __nonzero__
 
     @propertycache
     def _status(self):
         return self._repo.status()
 
     @propertycache
     def _user(self):
         return self._repo.ui.username()
 
     @propertycache
     def _date(self):
         ui = self._repo.ui
         date = ui.configdate(b'devel', b'default-date')
         if date is None:
             date = dateutil.makedate()
         return date
 
     def subrev(self, subpath):
         return None
 
     def manifestnode(self):
         return None
 
     def user(self):
         return self._user or self._repo.ui.username()
 
     def date(self):
         return self._date
 
     def description(self):
         return self._text
 
     def files(self):
         return sorted(
             self._status.modified + self._status.added + self._status.removed
         )
 
     def modified(self):
         return self._status.modified
 
     def added(self):
         return self._status.added
 
     def removed(self):
         return self._status.removed
 
     def deleted(self):
         return self._status.deleted
 
     filesmodified = modified
     filesadded = added
     filesremoved = removed
 
     def branch(self):
         return encoding.tolocal(self._extra[b'branch'])
 
     def closesbranch(self):
         return b'close' in self._extra
 
     def extra(self):
         return self._extra
 
     def isinmemory(self):
         return False
 
     def tags(self):
         return []
 
     def bookmarks(self):
         b = []
         for p in self.parents():
             b.extend(p.bookmarks())
         return b
 
     def phase(self):
         phase = phases.newcommitphase(self._repo.ui)
         for p in self.parents():
             phase = max(phase, p.phase())
         return phase
 
     def hidden(self):
         return False
 
     def children(self):
         return []
 
     def flags(self, path):
         if '_manifest' in self.__dict__:
             try:
                 return self._manifest.flags(path)
             except KeyError:
                 return b''
 
         try:
             return self._flagfunc(path)
         except OSError:
             return b''
 
     def ancestor(self, c2):
         """return the "best" ancestor context of self and c2"""
         return self._parents[0].ancestor(c2)  # punt on two parents for now
 
     def ancestors(self):
         for p in self._parents:
             yield p
         for a in self._repo.changelog.ancestors(
             [p.rev() for p in self._parents]
         ):
             yield self._repo[a]
 
     def markcommitted(self, node):
         """Perform post-commit cleanup necessary after committing this ctx
 
         Specifically, this updates backing stores this working context
         wraps to reflect the fact that the changes reflected by this
         workingctx have been committed.  For example, it marks
         modified and added files as normal in the dirstate.
 
         """
 
     def dirty(self, missing=False, merge=True, branch=True):
         return False
 
 
 class workingctx(committablectx):
     """A workingctx object makes access to data related to
     the current working directory convenient.
     date - any valid date string or (unixtime, offset), or None.
     user - username string, or None.
     extra - a dictionary of extra values, or None.
     changes - a list of file lists as returned by localrepo.status()
                or None to use the repository status.
     """
 
     def __init__(
         self, repo, text=b"", user=None, date=None, extra=None, changes=None
     ):
         branch = None
         if not extra or b'branch' not in extra:
             try:
                 branch = repo.dirstate.branch()
             except UnicodeDecodeError:
                 raise error.Abort(_(b'branch name not in UTF-8!'))
         super(workingctx, self).__init__(
             repo, text, user, date, extra, changes, branch=branch
         )
 
     def __iter__(self):
         d = self._repo.dirstate
         for f in d:
             if d.get_entry(f).tracked:
                 yield f
 
     def __contains__(self, key):
         return self._repo.dirstate.get_entry(key).tracked
 
     def hex(self):
         return self._repo.nodeconstants.wdirhex
 
     @propertycache
     def _parents(self):
         p = self._repo.dirstate.parents()
         if p[1] == self._repo.nodeconstants.nullid:
             p = p[:-1]
         # use unfiltered repo to delay/avoid loading obsmarkers
         unfi = self._repo.unfiltered()
         return [
             changectx(
                 self._repo, unfi.changelog.rev(n), n, maybe_filtered=False
             )
             for n in p
         ]
 
     def setparents(self, p1node, p2node=None):
         if p2node is None:
             p2node = self._repo.nodeconstants.nullid
         dirstate = self._repo.dirstate
         with dirstate.parentchange():
             copies = dirstate.setparents(p1node, p2node)
             pctx = self._repo[p1node]
             if copies:
                 # Adjust copy records, the dirstate cannot do it, it
                 # requires access to parents manifests. Preserve them
                 # only for entries added to first parent.
                 for f in copies:
                     if f not in pctx and copies[f] in pctx:
                         dirstate.copy(copies[f], f)
             if p2node == self._repo.nodeconstants.nullid:
                 for f, s in sorted(dirstate.copies().items()):
                     if f not in pctx and s not in pctx:
                         dirstate.copy(None, f)
 
     def _fileinfo(self, path):
         # populate __dict__['_manifest'] as workingctx has no _manifestdelta
         self._manifest
         return super(workingctx, self)._fileinfo(path)
 
     def _buildflagfunc(self):
         # Create a fallback function for getting file flags when the
         # filesystem doesn't support them
 
         copiesget = self._repo.dirstate.copies().get
         parents = self.parents()
         if len(parents) < 2:
             # when we have one parent, it's easy: copy from parent
             man = parents[0].manifest()
 
             def func(f):
                 f = copiesget(f, f)
                 return man.flags(f)
 
         else:
             # merges are tricky: we try to reconstruct the unstored
             # result from the merge (issue1802)
             p1, p2 = parents
             pa = p1.ancestor(p2)
             m1, m2, ma = p1.manifest(), p2.manifest(), pa.manifest()
 
             def func(f):
                 f = copiesget(f, f)  # may be wrong for merges with copies
                 fl1, fl2, fla = m1.flags(f), m2.flags(f), ma.flags(f)
                 if fl1 == fl2:
                     return fl1
                 if fl1 == fla:
                     return fl2
                 if fl2 == fla:
                     return fl1
                 return b''  # punt for conflicts
 
         return func
 
     @propertycache
     def _flagfunc(self):
         return self._repo.dirstate.flagfunc(self._buildflagfunc)
 
     def flags(self, path):
         try:
             return self._flagfunc(path)
         except OSError:
             return b''
 
     def filectx(self, path, filelog=None):
         """get a file context from the working directory"""
         return workingfilectx(
             self._repo, path, workingctx=self, filelog=filelog
         )
 
     def dirty(self, missing=False, merge=True, branch=True):
         """check whether a working directory is modified"""
         # check subrepos first
         for s in sorted(self.substate):
             if self.sub(s).dirty(missing=missing):
                 return True
         # check current working dir
         return (
             (merge and self.p2())
             or (branch and self.branch() != self.p1().branch())
             or self.modified()
             or self.added()
             or self.removed()
             or (missing and self.deleted())
         )
 
     def add(self, list, prefix=b""):
         with self._repo.wlock():
             ui, ds = self._repo.ui, self._repo.dirstate
             uipath = lambda f: ds.pathto(pathutil.join(prefix, f))
             rejected = []
             lstat = self._repo.wvfs.lstat
             for f in list:
                 # ds.pathto() returns an absolute file when this is invoked from
                 # the keyword extension.  That gets flagged as non-portable on
                 # Windows, since it contains the drive letter and colon.
                 scmutil.checkportable(ui, os.path.join(prefix, f))
                 try:
                     st = lstat(f)
                 except OSError:
                     ui.warn(_(b"%s does not exist!\n") % uipath(f))
                     rejected.append(f)
                     continue
                 limit = ui.configbytes(b'ui', b'large-file-limit')
                 if limit != 0 and st.st_size > limit:
                     ui.warn(
                         _(
                             b"%s: up to %d MB of RAM may be required "
                             b"to manage this file\n"
                             b"(use 'hg revert %s' to cancel the "
                             b"pending addition)\n"
                         )
                         % (f, 3 * st.st_size // 1000000, uipath(f))
                     )
                 if not (stat.S_ISREG(st.st_mode) or stat.S_ISLNK(st.st_mode)):
                     ui.warn(
                         _(
                             b"%s not added: only files and symlinks "
                             b"supported currently\n"
                         )
                         % uipath(f)
                     )
                     rejected.append(f)
                 elif not ds.set_tracked(f):
                     ui.warn(_(b"%s already tracked!\n") % uipath(f))
             return rejected
 
     def forget(self, files, prefix=b""):
         with self._repo.wlock():
             ds = self._repo.dirstate
             uipath = lambda f: ds.pathto(pathutil.join(prefix, f))
             rejected = []
             for f in files:
                 if not ds.set_untracked(f):
                     self._repo.ui.warn(_(b"%s not tracked!\n") % uipath(f))
                     rejected.append(f)
             return rejected
 
     def copy(self, source, dest):
         try:
             st = self._repo.wvfs.lstat(dest)
         except FileNotFoundError:
             self._repo.ui.warn(
                 _(b"%s does not exist!\n") % self._repo.dirstate.pathto(dest)
             )
             return
         if not (stat.S_ISREG(st.st_mode) or stat.S_ISLNK(st.st_mode)):
             self._repo.ui.warn(
                 _(b"copy failed: %s is not a file or a symbolic link\n")
                 % self._repo.dirstate.pathto(dest)
             )
         else:
             with self._repo.wlock():
                 ds = self._repo.dirstate
                 ds.set_tracked(dest)
                 ds.copy(source, dest)
 
     def match(
         self,
         pats=None,
         include=None,
         exclude=None,
         default=b'glob',
         listsubrepos=False,
         badfn=None,
         cwd=None,
     ):
         r = self._repo
         if not cwd:
             cwd = r.getcwd()
 
         # Only a case insensitive filesystem needs magic to translate user input
         # to actual case in the filesystem.
         icasefs = not util.fscasesensitive(r.root)
         return matchmod.match(
             r.root,
             cwd,
             pats,
             include,
             exclude,
             default,
             auditor=r.auditor,
             ctx=self,
             listsubrepos=listsubrepos,
             badfn=badfn,
             icasefs=icasefs,
         )
 
     def _filtersuspectsymlink(self, files):
         if not files or self._repo.dirstate._checklink:
             return files
 
         # Symlink placeholders may get non-symlink-like contents
         # via user error or dereferencing by NFS or Samba servers,
         # so we filter out any placeholders that don't look like a
         # symlink
         sane = []
         for f in files:
             if self.flags(f) == b'l':
                 d = self[f].data()
                 if (
                     d == b''
                     or len(d) >= 1024
                     or b'\n' in d
                     or stringutil.binary(d)
                 ):
                     self._repo.ui.debug(
                         b'ignoring suspect symlink placeholder "%s"\n' % f
                     )
                     continue
             sane.append(f)
         return sane
 
     def _checklookup(self, files, mtime_boundary):
         # check for any possibly clean files
         if not files:
             return [], [], [], []
 
         modified = []
         deleted = []
         clean = []
         fixup = []
         pctx = self._parents[0]
         # do a full compare of any files that might have changed
         for f in sorted(files):
             try:
                 # This will return True for a file that got replaced by a
                 # directory in the interim, but fixing that is pretty hard.
                 if (
                     f not in pctx
                     or self.flags(f) != pctx.flags(f)
                     or pctx[f].cmp(self[f])
                 ):
                     modified.append(f)
                 elif mtime_boundary is None:
                     clean.append(f)
                 else:
                     s = self[f].lstat()
                     mode = s.st_mode
                     size = s.st_size
                     file_mtime = timestamp.reliable_mtime_of(s, mtime_boundary)
                     if file_mtime is not None:
                         cache_info = (mode, size, file_mtime)
                         fixup.append((f, cache_info))
                     else:
                         clean.append(f)
             except (IOError, OSError):
                 # A file become inaccessible in between? Mark it as deleted,
                 # matching dirstate behavior (issue5584).
                 # The dirstate has more complex behavior around whether a
                 # missing file matches a directory, etc, but we don't need to
                 # bother with that: if f has made it to this point, we're sure
                 # it's in the dirstate.
                 deleted.append(f)
 
         return modified, deleted, clean, fixup
 
     def _poststatusfixup(self, status, fixup):
         """update dirstate for files that are actually clean"""
+        ui = self._repo.ui
+        testing.wait_on_cfg(self._repo.ui, b'status.pre-dirstate-write-file')
         poststatus = self._repo.postdsstatus()
         if fixup or 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
                 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'
-                        )
+                        ui.debug(b'skip updating dirstate: identity mismatch\n')
             except error.LockError:
                 pass
             finally:
                 # Even if the wlock couldn't be grabbed, clear out the list.
                 self._repo.clearpostdsstatus()
 
     def _dirstatestatus(self, match, ignored=False, clean=False, unknown=False):
         '''Gets the status from the dirstate -- internal use only.'''
         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
             )
             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
             if s.unknown or s.ignored or s.clean:
                 # "_status" is cached with list*=False in the normal route
                 self._status = scmutil.status(
                     s.modified, s.added, s.removed, s.deleted, [], [], []
                 )
             else:
                 self._status = s
 
         return s
 
     @propertycache
     def _copies(self):
         p1copies = {}
         p2copies = {}
         parents = self._repo.dirstate.parents()
         p1manifest = self._repo[parents[0]].manifest()
         p2manifest = self._repo[parents[1]].manifest()
         changedset = set(self.added()) | set(self.modified())
         narrowmatch = self._repo.narrowmatch()
         for dst, src in self._repo.dirstate.copies().items():
             if dst not in changedset or not narrowmatch(dst):
                 continue
             if src in p1manifest:
                 p1copies[dst] = src
             elif src in p2manifest:
                 p2copies[dst] = src
         return p1copies, p2copies
 
     @propertycache
     def _manifest(self):
         """generate a manifest corresponding to the values in self._status
 
         This reuse the file nodeid from parent, but we use special node
         identifiers for added and modified files. This is used by manifests
         merge to see that files are different and by update logic to avoid
         deleting newly added files.
         """
         return self._buildstatusmanifest(self._status)
 
     def _buildstatusmanifest(self, status):
         """Builds a manifest that includes the given status results."""
         parents = self.parents()
 
         man = parents[0].manifest().copy()
 
         ff = self._flagfunc
         for i, l in (
             (self._repo.nodeconstants.addednodeid, status.added),
             (self._repo.nodeconstants.modifiednodeid, status.modified),
         ):
             for f in l:
                 man[f] = i
                 try:
                     man.setflag(f, ff(f))
                 except OSError:
                     pass
 
         for f in status.deleted + status.removed:
             if f in man:
                 del man[f]
 
         return man
 
     def _buildstatus(
         self, other, s, match, listignored, listclean, listunknown
     ):
         """build a status with respect to another context
 
         This includes logic for maintaining the fast path of status when
         comparing the working directory against its parent, which is to skip
         building a new manifest if self (working directory) is not comparing
         against its parent (repo['.']).
         """
         s = self._dirstatestatus(match, listignored, listclean, listunknown)
         # Filter out symlinks that, in the case of FAT32 and NTFS filesystems,
         # might have accidentally ended up with the entire contents of the file
         # they are supposed to be linking to.
         s.modified[:] = self._filtersuspectsymlink(s.modified)
         if other != self._repo[b'.']:
             s = super(workingctx, self)._buildstatus(
                 other, s, match, listignored, listclean, listunknown
             )
         return s
 
     def _matchstatus(self, other, match):
         """override the match method with a filter for directory patterns
 
         We use inheritance to customize the match.bad method only in cases of
         workingctx since it belongs only to the working directory when
         comparing against the parent changeset.
 
         If we aren't comparing against the working directory's parent, then we
         just use the default match object sent to us.
         """
         if other != self._repo[b'.']:
 
             def bad(f, msg):
                 # 'f' may be a directory pattern from 'match.files()',
                 # so 'f not in ctx1' is not enough
                 if f not in other and not other.hasdir(f):
                     self._repo.ui.warn(
                         b'%s: %s\n' % (self._repo.dirstate.pathto(f), msg)
                     )
 
             match.bad = bad
         return match
 
     def walk(self, match):
         '''Generates matching file names.'''
         return sorted(
             self._repo.dirstate.walk(
                 self._repo.narrowmatch(match),
                 subrepos=sorted(self.substate),
                 unknown=True,
                 ignored=False,
             )
         )
 
     def matches(self, match):
         match = self._repo.narrowmatch(match)
         ds = self._repo.dirstate
         return sorted(f for f in ds.matches(match) if ds.get_entry(f).tracked)
 
     def markcommitted(self, node):
         with self._repo.dirstate.parentchange():
             for f in self.modified() + self.added():
                 self._repo.dirstate.update_file(
                     f, p1_tracked=True, wc_tracked=True
                 )
             for f in self.removed():
                 self._repo.dirstate.update_file(
                     f, p1_tracked=False, wc_tracked=False
                 )
             self._repo.dirstate.setparents(node)
             self._repo._quick_access_changeid_invalidate()
 
             sparse.aftercommit(self._repo, node)
 
         # write changes out explicitly, because nesting wlock at
         # runtime may prevent 'wlock.release()' in 'repo.commit()'
         # from immediately doing so for subsequent changing files
         self._repo.dirstate.write(self._repo.currenttransaction())
 
     def mergestate(self, clean=False):
         if clean:
             return mergestatemod.mergestate.clean(self._repo)
         return mergestatemod.mergestate.read(self._repo)
 
 
 class committablefilectx(basefilectx):
     """A committablefilectx provides common functionality for a file context
     that wants the ability to commit, e.g. workingfilectx or memfilectx."""
 
     def __init__(self, repo, path, filelog=None, ctx=None):
         self._repo = repo
         self._path = path
         self._changeid = None
         self._filerev = self._filenode = None
 
         if filelog is not None:
             self._filelog = filelog
         if ctx:
             self._changectx = ctx
 
     def __nonzero__(self):
         return True
 
     __bool__ = __nonzero__
 
     def linkrev(self):
         # linked to self._changectx no matter if file is modified or not
         return self.rev()
 
     def renamed(self):
         path = self.copysource()
         if not path:
             return None
         return (
             path,
             self._changectx._parents[0]._manifest.get(
                 path, self._repo.nodeconstants.nullid
             ),
         )
 
     def parents(self):
         '''return parent filectxs, following copies if necessary'''
 
         def filenode(ctx, path):
             return ctx._manifest.get(path, self._repo.nodeconstants.nullid)
 
         path = self._path
         fl = self._filelog
         pcl = self._changectx._parents
         renamed = self.renamed()
 
         if renamed:
             pl = [renamed + (None,)]
         else:
             pl = [(path, filenode(pcl[0], path), fl)]
 
         for pc in pcl[1:]:
             pl.append((path, filenode(pc, path), fl))
 
         return [
             self._parentfilectx(p, fileid=n, filelog=l)
             for p, n, l in pl
             if n != self._repo.nodeconstants.nullid
         ]
 
     def children(self):
         return []
 
 
 class workingfilectx(committablefilectx):
     """A workingfilectx object makes access to data related to a particular
     file in the working directory convenient."""
 
     def __init__(self, repo, path, filelog=None, workingctx=None):
         super(workingfilectx, self).__init__(repo, path, filelog, workingctx)
 
     @propertycache
     def _changectx(self):
         return workingctx(self._repo)
 
     def data(self):
         return self._repo.wread(self._path)
 
     def copysource(self):
         return self._repo.dirstate.copied(self._path)
 
     def size(self):
         return self._repo.wvfs.lstat(self._path).st_size
 
     def lstat(self):
         return self._repo.wvfs.lstat(self._path)
 
     def date(self):
         t, tz = self._changectx.date()
         try:
             return (self._repo.wvfs.lstat(self._path)[stat.ST_MTIME], tz)
         except FileNotFoundError:
             return (t, tz)
 
     def exists(self):
         return self._repo.wvfs.exists(self._path)
 
     def lexists(self):
         return self._repo.wvfs.lexists(self._path)
 
     def audit(self):
         return self._repo.wvfs.audit(self._path)
 
     def cmp(self, fctx):
         """compare with other file context
 
         returns True if different than fctx.
         """
         # fctx should be a filectx (not a workingfilectx)
         # invert comparison to reuse the same code path
         return fctx.cmp(self)
 
     def remove(self, ignoremissing=False):
         """wraps unlink for a repo's working directory"""
         rmdir = self._repo.ui.configbool(b'experimental', b'removeemptydirs')
         self._repo.wvfs.unlinkpath(
             self._path, ignoremissing=ignoremissing, rmdir=rmdir
         )
 
     def write(self, data, flags, backgroundclose=False, **kwargs):
         """wraps repo.wwrite"""
         return self._repo.wwrite(
             self._path, data, flags, backgroundclose=backgroundclose, **kwargs
         )
 
     def markcopied(self, src):
         """marks this file a copy of `src`"""
         self._repo.dirstate.copy(src, self._path)
 
     def clearunknown(self):
         """Removes conflicting items in the working directory so that
         ``write()`` can be called successfully.
         """
         wvfs = self._repo.wvfs
         f = self._path
         wvfs.audit(f)
         if self._repo.ui.configbool(
             b'experimental', b'merge.checkpathconflicts'
         ):
             # remove files under the directory as they should already be
             # warned and backed up
             if wvfs.isdir(f) and not wvfs.islink(f):
                 wvfs.rmtree(f, forcibly=True)
             for p in reversed(list(pathutil.finddirs(f))):
                 if wvfs.isfileorlink(p):
                     wvfs.unlink(p)
                     break
         else:
             # don't remove files if path conflicts are not processed
             if wvfs.isdir(f) and not wvfs.islink(f):
                 wvfs.removedirs(f)
 
     def setflags(self, l, x):
         self._repo.wvfs.setflags(self._path, l, x)
 
 
 class overlayworkingctx(committablectx):
     """Wraps another mutable context with a write-back cache that can be
     converted into a commit context.
 
     self._cache[path] maps to a dict with keys: {
         'exists': bool?
         'date': date?
         'data': str?
         'flags': str?
         'copied': str? (path or None)
     }
     If `exists` is True, `flags` must be non-None and 'date' is non-None. If it
     is `False`, the file was deleted.
     """
 
     def __init__(self, repo):
         super(overlayworkingctx, self).__init__(repo)
         self.clean()
 
     def setbase(self, wrappedctx):
         self._wrappedctx = wrappedctx
         self._parents = [wrappedctx]
         # Drop old manifest cache as it is now out of date.
         # This is necessary when, e.g., rebasing several nodes with one
         # ``overlayworkingctx`` (e.g. with --collapse).
         util.clearcachedproperty(self, b'_manifest')
 
     def setparents(self, p1node, p2node=None):
         if p2node is None:
             p2node = self._repo.nodeconstants.nullid
         assert p1node == self._wrappedctx.node()
         self._parents = [self._wrappedctx, self._repo.unfiltered()[p2node]]
 
     def data(self, path):
         if self.isdirty(path):
             if self._cache[path][b'exists']:
                 if self._cache[path][b'data'] is not None:
                     return self._cache[path][b'data']
                 else:
                     # Must fallback here, too, because we only set flags.
                     return self._wrappedctx[path].data()
             else:
                 raise error.ProgrammingError(
                     b"No such file or directory: %s" % path
                 )
         else:
             return self._wrappedctx[path].data()
 
     @propertycache
     def _manifest(self):
         parents = self.parents()
         man = parents[0].manifest().copy()
 
         flag = self._flagfunc
         for path in self.added():
             man[path] = self._repo.nodeconstants.addednodeid
             man.setflag(path, flag(path))
         for path in self.modified():
             man[path] = self._repo.nodeconstants.modifiednodeid
             man.setflag(path, flag(path))
         for path in self.removed():
             del man[path]
         return man
 
     @propertycache
     def _flagfunc(self):
         def f(path):
             return self._cache[path][b'flags']
 
         return f
 
     def files(self):
         return sorted(self.added() + self.modified() + self.removed())
 
     def modified(self):
         return [
             f
             for f in self._cache.keys()
             if self._cache[f][b'exists'] and self._existsinparent(f)
         ]
 
     def added(self):
         return [
             f
             for f in self._cache.keys()
             if self._cache[f][b'exists'] and not self._existsinparent(f)
         ]
 
     def removed(self):
         return [
             f
             for f in self._cache.keys()
             if not self._cache[f][b'exists'] and self._existsinparent(f)
         ]
 
     def p1copies(self):
         copies = {}
         narrowmatch = self._repo.narrowmatch()
         for f in self._cache.keys():
             if not narrowmatch(f):
                 continue
             copies.pop(f, None)  # delete if it exists
             source = self._cache[f][b'copied']
             if source:
                 copies[f] = source
         return copies
 
     def p2copies(self):
         copies = {}
         narrowmatch = self._repo.narrowmatch()
         for f in self._cache.keys():
             if not narrowmatch(f):
                 continue
             copies.pop(f, None)  # delete if it exists
             source = self._cache[f][b'copied']
             if source:
                 copies[f] = source
         return copies
 
     def isinmemory(self):
         return True
 
     def filedate(self, path):
         if self.isdirty(path):
             return self._cache[path][b'date']
         else:
             return self._wrappedctx[path].date()
 
     def markcopied(self, path, origin):
         self._markdirty(
             path,
             exists=True,
             date=self.filedate(path),
             flags=self.flags(path),
             copied=origin,
         )
 
     def copydata(self, path):
         if self.isdirty(path):
             return self._cache[path][b'copied']
         else:
             return None
 
     def flags(self, path):
         if self.isdirty(path):
             if self._cache[path][b'exists']:
                 return self._cache[path][b'flags']
             else:
                 raise error.ProgrammingError(
                     b"No such file or directory: %s" % path
                 )
         else:
             return self._wrappedctx[path].flags()
 
     def __contains__(self, key):
         if key in self._cache:
             return self._cache[key][b'exists']
         return key in self.p1()
 
     def _existsinparent(self, path):
         try:
             # ``commitctx` raises a ``ManifestLookupError`` if a path does not
             # exist, unlike ``workingctx``, which returns a ``workingfilectx``
             # with an ``exists()`` function.
             self._wrappedctx[path]
             return True
         except error.ManifestLookupError:
             return False
 
     def _auditconflicts(self, path):
         """Replicates conflict checks done by wvfs.write().
 
         Since we never write to the filesystem and never call `applyupdates` in
         IMM, we'll never check that a path is actually writable -- e.g., because
         it adds `a/foo`, but `a` is actually a file in the other commit.
         """
 
         def fail(path, component):
             # p1() is the base and we're receiving "writes" for p2()'s
             # files.
             if b'l' in self.p1()[component].flags():
                 raise error.Abort(
                     b"error: %s conflicts with symlink %s "
                     b"in %d." % (path, component, self.p1().rev())
                 )
             else:
                 raise error.Abort(
                     b"error: '%s' conflicts with file '%s' in "
                     b"%d." % (path, component, self.p1().rev())
                 )
 
         # Test that each new directory to be created to write this path from p2
         # is not a file in p1.
         components = path.split(b'/')
         for i in range(len(components)):
             component = b"/".join(components[0:i])
             if component in self:
                 fail(path, component)
 
         # Test the other direction -- that this path from p2 isn't a directory
         # in p1 (test that p1 doesn't have any paths matching `path/*`).
         match = self.match([path], default=b'path')
         mfiles = list(self.p1().manifest().walk(match))
         if len(mfiles) > 0:
             if len(mfiles) == 1 and mfiles[0] == path:
                 return
             # omit the files which are deleted in current IMM wctx
             mfiles = [m for m in mfiles if m in self]
             if not mfiles:
                 return
             raise error.Abort(
                 b"error: file '%s' cannot be written because "
                 b" '%s/' is a directory in %s (containing %d "
                 b"entries: %s)"
                 % (path, path, self.p1(), len(mfiles), b', '.join(mfiles))
             )
 
     def write(self, path, data, flags=b'', **kwargs):
         if data is None:
             raise error.ProgrammingError(b"data must be non-None")
         self._auditconflicts(path)
         self._markdirty(
             path, exists=True, data=data, date=dateutil.makedate(), flags=flags
         )
 
     def setflags(self, path, l, x):
         flag = b''
         if l:
             flag = b'l'
         elif x:
             flag = b'x'
         self._markdirty(path, exists=True, date=dateutil.makedate(), flags=flag)
 
     def remove(self, path):
         self._markdirty(path, exists=False)
 
     def exists(self, path):
         """exists behaves like `lexists`, but needs to follow symlinks and
         return False if they are broken.
         """
         if self.isdirty(path):
             # If this path exists and is a symlink, "follow" it by calling
             # exists on the destination path.
             if (
                 self._cache[path][b'exists']
                 and b'l' in self._cache[path][b'flags']
             ):
                 return self.exists(self._cache[path][b'data'].strip())
             else:
                 return self._cache[path][b'exists']
 
         return self._existsinparent(path)
 
     def lexists(self, path):
         """lexists returns True if the path exists"""
         if self.isdirty(path):
             return self._cache[path][b'exists']
 
         return self._existsinparent(path)
 
     def size(self, path):
         if self.isdirty(path):
             if self._cache[path][b'exists']:
                 return len(self._cache[path][b'data'])
             else:
                 raise error.ProgrammingError(
                     b"No such file or directory: %s" % path
                 )
         return self._wrappedctx[path].size()
 
     def tomemctx(
         self,
         text,
         branch=None,
         extra=None,
         date=None,
         parents=None,
         user=None,
         editor=None,
     ):
         """Converts this ``overlayworkingctx`` into a ``memctx`` ready to be
         committed.
 
         ``text`` is the commit message.
         ``parents`` (optional) are rev numbers.
         """
         # Default parents to the wrapped context if not passed.
         if parents is None:
             parents = self.parents()
             if len(parents) == 1:
                 parents = (parents[0], None)
 
         # ``parents`` is passed as rev numbers; convert to ``commitctxs``.
         if parents[1] is None:
             parents = (self._repo[parents[0]], None)
         else:
             parents = (self._repo[parents[0]], self._repo[parents[1]])
 
         files = self.files()
 
         def getfile(repo, memctx, path):
             if self._cache[path][b'exists']:
                 return memfilectx(
                     repo,
                     memctx,
                     path,
                     self._cache[path][b'data'],
                     b'l' in self._cache[path][b'flags'],
                     b'x' in self._cache[path][b'flags'],
                     self._cache[path][b'copied'],
                 )
             else:
                 # Returning None, but including the path in `files`, is
                 # necessary for memctx to register a deletion.
                 return None
 
         if branch is None:
             branch = self._wrappedctx.branch()
 
         return memctx(
             self._repo,
             parents,
             text,
             files,
             getfile,
             date=date,
             extra=extra,
             user=user,
             branch=branch,
             editor=editor,
         )
 
     def tomemctx_for_amend(self, precursor):
         extra = precursor.extra().copy()
         extra[b'amend_source'] = precursor.hex()
         return self.tomemctx(
             text=precursor.description(),
             branch=precursor.branch(),
             extra=extra,
             date=precursor.date(),
             user=precursor.user(),
         )
 
     def isdirty(self, path):
         return path in self._cache
 
     def clean(self):
         self._mergestate = None
         self._cache = {}
 
     def _compact(self):
         """Removes keys from the cache that are actually clean, by comparing
         them with the underlying context.
 
         This can occur during the merge process, e.g. by passing --tool :local
         to resolve a conflict.
         """
         keys = []
         # This won't be perfect, but can help performance significantly when
         # using things like remotefilelog.
         scmutil.prefetchfiles(
             self.repo(),
             [
                 (
                     self.p1().rev(),
                     scmutil.matchfiles(self.repo(), self._cache.keys()),
                 )
             ],
         )
 
         for path in self._cache.keys():
             cache = self._cache[path]
             try:
                 underlying = self._wrappedctx[path]
                 if (
                     underlying.data() == cache[b'data']
                     and underlying.flags() == cache[b'flags']
                 ):
                     keys.append(path)
             except error.ManifestLookupError:
                 # Path not in the underlying manifest (created).
                 continue
 
         for path in keys:
             del self._cache[path]
         return keys
 
     def _markdirty(
         self, path, exists, data=None, date=None, flags=b'', copied=None
     ):
         # data not provided, let's see if we already have some; if not, let's
         # grab it from our underlying context, so that we always have data if
         # the file is marked as existing.
         if exists and data is None:
             oldentry = self._cache.get(path) or {}
             data = oldentry.get(b'data')
             if data is None:
                 data = self._wrappedctx[path].data()
 
         self._cache[path] = {
             b'exists': exists,
             b'data': data,
             b'date': date,
             b'flags': flags,
             b'copied': copied,
         }
         util.clearcachedproperty(self, b'_manifest')
 
     def filectx(self, path, filelog=None):
         return overlayworkingfilectx(
             self._repo, path, parent=self, filelog=filelog
         )
 
     def mergestate(self, clean=False):
         if clean or self._mergestate is None:
             self._mergestate = mergestatemod.memmergestate(self._repo)
         return self._mergestate
 
 
 class overlayworkingfilectx(committablefilectx):
     """Wrap a ``workingfilectx`` but intercepts all writes into an in-memory
     cache, which can be flushed through later by calling ``flush()``."""
 
     def __init__(self, repo, path, filelog=None, parent=None):
         super(overlayworkingfilectx, self).__init__(repo, path, filelog, parent)
         self._repo = repo
         self._parent = parent
         self._path = path
 
     def cmp(self, fctx):
         return self.data() != fctx.data()
 
     def changectx(self):
         return self._parent
 
     def data(self):
         return self._parent.data(self._path)
 
     def date(self):
         return self._parent.filedate(self._path)
 
     def exists(self):
         return self.lexists()
 
     def lexists(self):
         return self._parent.exists(self._path)
 
     def copysource(self):
         return self._parent.copydata(self._path)
 
     def size(self):
         return self._parent.size(self._path)
 
     def markcopied(self, origin):
         self._parent.markcopied(self._path, origin)
 
     def audit(self):
         pass
 
     def flags(self):
         return self._parent.flags(self._path)
 
     def setflags(self, islink, isexec):
         return self._parent.setflags(self._path, islink, isexec)
 
     def write(self, data, flags, backgroundclose=False, **kwargs):
         return self._parent.write(self._path, data, flags, **kwargs)
 
     def remove(self, ignoremissing=False):
         return self._parent.remove(self._path)
 
     def clearunknown(self):
         pass
 
 
 class workingcommitctx(workingctx):
     """A workingcommitctx object makes access to data related to
     the revision being committed convenient.
 
     This hides changes in the working directory, if they aren't
     committed in this context.
     """
 
     def __init__(
         self, repo, changes, text=b"", user=None, date=None, extra=None
     ):
         super(workingcommitctx, self).__init__(
             repo, text, user, date, extra, changes
         )
 
     def _dirstatestatus(self, match, ignored=False, clean=False, unknown=False):
         """Return matched files only in ``self._status``
 
         Uncommitted files appear "clean" via this context, even if
         they aren't actually so in the working directory.
         """
         if clean:
             clean = [f for f in self._manifest if f not in self._changedset]
         else:
             clean = []
         return scmutil.status(
             [f for f in self._status.modified if match(f)],
             [f for f in self._status.added if match(f)],
             [f for f in self._status.removed if match(f)],
             [],
             [],
             [],
             clean,
         )
 
     @propertycache
     def _changedset(self):
         """Return the set of files changed in this context"""
         changed = set(self._status.modified)
         changed.update(self._status.added)
         changed.update(self._status.removed)
         return changed
 
 
 def makecachingfilectxfn(func):
     """Create a filectxfn that caches based on the path.
 
     We can't use util.cachefunc because it uses all arguments as the cache
     key and this creates a cycle since the arguments include the repo and
     memctx.
     """
     cache = {}
 
     def getfilectx(repo, memctx, path):
         if path not in cache:
             cache[path] = func(repo, memctx, path)
         return cache[path]
 
     return getfilectx
 
 
 def memfilefromctx(ctx):
     """Given a context return a memfilectx for ctx[path]
 
     This is a convenience method for building a memctx based on another
     context.
     """
 
     def getfilectx(repo, memctx, path):
         fctx = ctx[path]
         copysource = fctx.copysource()
         return memfilectx(
             repo,
             memctx,
             path,
             fctx.data(),
             islink=fctx.islink(),
             isexec=fctx.isexec(),
             copysource=copysource,
         )
 
     return getfilectx
 
 
 def memfilefrompatch(patchstore):
     """Given a patch (e.g. patchstore object) return a memfilectx
 
     This is a convenience method for building a memctx based on a patchstore.
     """
 
     def getfilectx(repo, memctx, path):
         data, mode, copysource = patchstore.getfile(path)
         if data is None:
             return None
         islink, isexec = mode
         return memfilectx(
             repo,
             memctx,
             path,
             data,
             islink=islink,
             isexec=isexec,
             copysource=copysource,
         )
 
     return getfilectx
 
 
 class memctx(committablectx):
     """Use memctx to perform in-memory commits via localrepo.commitctx().
 
     Revision information is supplied at initialization time while
     related files data and is made available through a callback
     mechanism.  'repo' is the current localrepo, 'parents' is a
     sequence of two parent revisions identifiers (pass None for every
     missing parent), 'text' is the commit message and 'files' lists
     names of files touched by the revision (normalized and relative to
     repository root).
 
     filectxfn(repo, memctx, path) is a callable receiving the
     repository, the current memctx object and the normalized path of
     requested file, relative to repository root. It is fired by the
     commit function for every file in 'files', but calls order is
     undefined. If the file is available in the revision being
     committed (updated or added), filectxfn returns a memfilectx
     object. If the file was removed, filectxfn return None for recent
     Mercurial. Moved files are represented by marking the source file
     removed and the new file added with copy information (see
     memfilectx).
 
     user receives the committer name and defaults to current
     repository username, date is the commit date in any format
     supported by dateutil.parsedate() and defaults to current date, extra
     is a dictionary of metadata or is left empty.
     """
 
     # Mercurial <= 3.1 expects the filectxfn to raise IOError for missing files.
     # Extensions that need to retain compatibility across Mercurial 3.1 can use
     # this field to determine what to do in filectxfn.
     _returnnoneformissingfiles = True
 
     def __init__(
         self,
         repo,
         parents,
         text,
         files,
         filectxfn,
         user=None,
         date=None,
         extra=None,
         branch=None,
         editor=None,
     ):
         super(memctx, self).__init__(
             repo, text, user, date, extra, branch=branch
         )
         self._rev = None
         self._node = None
         parents = [(p or self._repo.nodeconstants.nullid) for p in parents]
         p1, p2 = parents
         self._parents = [self._repo[p] for p in (p1, p2)]
         files = sorted(set(files))
         self._files = files
         self.substate = {}
 
         if isinstance(filectxfn, patch.filestore):
             filectxfn = memfilefrompatch(filectxfn)
         elif not callable(filectxfn):
             # if store is not callable, wrap it in a function
             filectxfn = memfilefromctx(filectxfn)
 
         # memoizing increases performance for e.g. vcs convert scenarios.
         self._filectxfn = makecachingfilectxfn(filectxfn)
 
         if editor:
             self._text = editor(self._repo, self, [])
             self._repo.savecommitmessage(self._text)
 
     def filectx(self, path, filelog=None):
         """get a file context from the working directory
 
         Returns None if file doesn't exist and should be removed."""
         return self._filectxfn(self._repo, self, path)
 
     def commit(self):
         """commit context to the repo"""
         return self._repo.commitctx(self)
 
     @propertycache
     def _manifest(self):
         """generate a manifest based on the return values of filectxfn"""
 
         # keep this simple for now; just worry about p1
         pctx = self._parents[0]
         man = pctx.manifest().copy()
 
         for f in self._status.modified:
             man[f] = self._repo.nodeconstants.modifiednodeid
 
         for f in self._status.added:
             man[f] = self._repo.nodeconstants.addednodeid
 
         for f in self._status.removed:
             if f in man:
                 del man[f]
 
         return man
 
     @propertycache
     def _status(self):
         """Calculate exact status from ``files`` specified at construction"""
         man1 = self.p1().manifest()
         p2 = self._parents[1]
         # "1 < len(self._parents)" can't be used for checking
         # existence of the 2nd parent, because "memctx._parents" is
         # explicitly initialized by the list, of which length is 2.
         if p2.rev() != nullrev:
             man2 = p2.manifest()
             managing = lambda f: f in man1 or f in man2
         else:
             managing = lambda f: f in man1
 
         modified, added, removed = [], [], []
         for f in self._files:
             if not managing(f):
                 added.append(f)
             elif self[f]:
                 modified.append(f)
             else:
                 removed.append(f)
 
         return scmutil.status(modified, added, removed, [], [], [], [])
 
     def parents(self):
         if self._parents[1].rev() == nullrev:
             return [self._parents[0]]
         return self._parents
 
 
 class memfilectx(committablefilectx):
     """memfilectx represents an in-memory file to commit.
 
     See memctx and committablefilectx for more details.
     """
 
     def __init__(
         self,
         repo,
         changectx,
         path,
         data,
         islink=False,
         isexec=False,
         copysource=None,
     ):
         """
         path is the normalized file path relative to repository root.
         data is the file content as a string.
         islink is True if the file is a symbolic link.
         isexec is True if the file is executable.
         copied is the source file path if current file was copied in the
         revision being committed, or None."""
         super(memfilectx, self).__init__(repo, path, None, changectx)
         self._data = data
         if islink:
             self._flags = b'l'
         elif isexec:
             self._flags = b'x'
         else:
             self._flags = b''
         self._copysource = copysource
 
     def copysource(self):
         return self._copysource
 
     def cmp(self, fctx):
         return self.data() != fctx.data()
 
     def data(self):
         return self._data
 
     def remove(self, ignoremissing=False):
         """wraps unlink for a repo's working directory"""
         # need to figure out what to do here
         del self._changectx[self._path]
 
     def write(self, data, flags, **kwargs):
         """wraps repo.wwrite"""
         self._data = data
 
 
 class metadataonlyctx(committablectx):
     """Like memctx but it's reusing the manifest of different commit.
     Intended to be used by lightweight operations that are creating
     metadata-only changes.
 
     Revision information is supplied at initialization time.  'repo' is the
     current localrepo, 'ctx' is original revision which manifest we're reuisng
     'parents' is a sequence of two parent revisions identifiers (pass None for
     every missing parent), 'text' is the commit.
 
     user receives the committer name and defaults to current repository
     username, date is the commit date in any format supported by
     dateutil.parsedate() and defaults to current date, extra is a dictionary of
     metadata or is left empty.
     """
 
     def __init__(
         self,
         repo,
         originalctx,
         parents=None,
         text=None,
         user=None,
         date=None,
         extra=None,
         editor=None,
     ):
         if text is None:
             text = originalctx.description()
         super(metadataonlyctx, self).__init__(repo, text, user, date, extra)
         self._rev = None
         self._node = None
         self._originalctx = originalctx
         self._manifestnode = originalctx.manifestnode()
         if parents is None:
             parents = originalctx.parents()
         else:
             parents = [repo[p] for p in parents if p is not None]
         parents = parents[:]
         while len(parents) < 2:
             parents.append(repo[nullrev])
         p1, p2 = self._parents = parents
 
         # sanity check to ensure that the reused manifest parents are
         # manifests of our commit parents
         mp1, mp2 = self.manifestctx().parents
         if p1 != self._repo.nodeconstants.nullid and p1.manifestnode() != mp1:
             raise RuntimeError(
                 r"can't reuse the manifest: its p1 "
                 r"doesn't match the new ctx p1"
             )
         if p2 != self._repo.nodeconstants.nullid and p2.manifestnode() != mp2:
             raise RuntimeError(
                 r"can't reuse the manifest: "
                 r"its p2 doesn't match the new ctx p2"
             )
 
         self._files = originalctx.files()
         self.substate = {}
 
         if editor:
             self._text = editor(self._repo, self, [])
             self._repo.savecommitmessage(self._text)
 
     def manifestnode(self):
         return self._manifestnode
 
     @property
     def _manifestctx(self):
         return self._repo.manifestlog[self._manifestnode]
 
     def filectx(self, path, filelog=None):
         return self._originalctx.filectx(path, filelog=filelog)
 
     def commit(self):
         """commit context to the repo"""
         return self._repo.commitctx(self)
 
     @property
     def _manifest(self):
         return self._originalctx.manifest()
 
     @propertycache
     def _status(self):
         """Calculate exact status from ``files`` specified in the ``origctx``
         and parents manifests.
         """
         man1 = self.p1().manifest()
         p2 = self._parents[1]
         # "1 < len(self._parents)" can't be used for checking
         # existence of the 2nd parent, because "metadataonlyctx._parents" is
         # explicitly initialized by the list, of which length is 2.
         if p2.rev() != nullrev:
             man2 = p2.manifest()
             managing = lambda f: f in man1 or f in man2
         else:
             managing = lambda f: f in man1
 
         modified, added, removed = [], [], []
         for f in self._files:
             if not managing(f):
                 added.append(f)
             elif f in self:
                 modified.append(f)
             else:
                 removed.append(f)
 
         return scmutil.status(modified, added, removed, [], [], [], [])
 
 
 class arbitraryfilectx:
     """Allows you to use filectx-like functions on a file in an arbitrary
     location on disk, possibly not in the working directory.
     """
 
     def __init__(self, path, repo=None):
         # Repo is optional because contrib/simplemerge uses this class.
         self._repo = repo
         self._path = path
 
     def cmp(self, fctx):
         # filecmp follows symlinks whereas `cmp` should not, so skip the fast
         # path if either side is a symlink.
         symlinks = b'l' in self.flags() or b'l' in fctx.flags()
         if not symlinks and isinstance(fctx, workingfilectx) and self._repo:
             # Add a fast-path for merge if both sides are disk-backed.
             # Note that filecmp uses the opposite return values (True if same)
             # from our cmp functions (True if different).
             return not filecmp.cmp(self.path(), self._repo.wjoin(fctx.path()))
         return self.data() != fctx.data()
 
     def path(self):
         return self._path
 
     def flags(self):
         return b''
 
     def data(self):
         return util.readfile(self._path)
 
     def decodeddata(self):
         return util.readfile(self._path)
 
     def remove(self):
         util.unlink(self._path)
 
     def write(self, data, flags, **kwargs):
         assert not flags
         util.writefile(self._path, data)

rust/rhg/src/commands/status.rs

@@ -1,626 +1,634 b''
 // status.rs
 //
 // Copyright 2020, Georges Racinet <[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.
 
 use crate::error::CommandError;
 use crate::ui::Ui;
 use crate::utils::path_utils::RelativizePaths;
 use clap::{Arg, SubCommand};
 use format_bytes::format_bytes;
 use hg::config::Config;
 use hg::dirstate::has_exec_bit;
 use hg::dirstate::status::StatusPath;
 use hg::dirstate::TruncatedTimestamp;
 use hg::errors::{HgError, IoResultExt};
 use hg::lock::LockError;
 use hg::manifest::Manifest;
 use hg::matchers::{AlwaysMatcher, IntersectionMatcher};
 use hg::repo::Repo;
+use hg::utils::debug::debug_wait_for_file;
 use hg::utils::files::get_bytes_from_os_string;
 use hg::utils::files::get_bytes_from_path;
 use hg::utils::files::get_path_from_bytes;
 use hg::utils::hg_path::{hg_path_to_path_buf, HgPath};
 use hg::DirstateStatus;
 use hg::PatternFileWarning;
 use hg::StatusError;
 use hg::StatusOptions;
 use hg::{self, narrow, sparse};
 use log::info;
 use rayon::prelude::*;
 use std::io;
 use std::path::PathBuf;
 
 pub const HELP_TEXT: &str = "
 Show changed files in the working directory
 
 This is a pure Rust version of `hg status`.
 
 Some options might be missing, check the list below.
 ";
 
 pub fn args() -> clap::App<'static, 'static> {
     SubCommand::with_name("status")
         .alias("st")
         .about(HELP_TEXT)
         .arg(
             Arg::with_name("all")
                 .help("show status of all files")
                 .short("-A")
                 .long("--all"),
         )
         .arg(
             Arg::with_name("modified")
                 .help("show only modified files")
                 .short("-m")
                 .long("--modified"),
         )
         .arg(
             Arg::with_name("added")
                 .help("show only added files")
                 .short("-a")
                 .long("--added"),
         )
         .arg(
             Arg::with_name("removed")
                 .help("show only removed files")
                 .short("-r")
                 .long("--removed"),
         )
         .arg(
             Arg::with_name("clean")
                 .help("show only clean files")
                 .short("-c")
                 .long("--clean"),
         )
         .arg(
             Arg::with_name("deleted")
                 .help("show only deleted files")
                 .short("-d")
                 .long("--deleted"),
         )
         .arg(
             Arg::with_name("unknown")
                 .help("show only unknown (not tracked) files")
                 .short("-u")
                 .long("--unknown"),
         )
         .arg(
             Arg::with_name("ignored")
                 .help("show only ignored files")
                 .short("-i")
                 .long("--ignored"),
         )
         .arg(
             Arg::with_name("copies")
                 .help("show source of copied files (DEFAULT: ui.statuscopies)")
                 .short("-C")
                 .long("--copies"),
         )
         .arg(
             Arg::with_name("no-status")
                 .help("hide status prefix")
                 .short("-n")
                 .long("--no-status"),
         )
         .arg(
             Arg::with_name("verbose")
                 .help("enable additional output")
                 .short("-v")
                 .long("--verbose"),
         )
 }
 
 /// Pure data type allowing the caller to specify file states to display
 #[derive(Copy, Clone, Debug)]
 pub struct DisplayStates {
     pub modified: bool,
     pub added: bool,
     pub removed: bool,
     pub clean: bool,
     pub deleted: bool,
     pub unknown: bool,
     pub ignored: bool,
 }
 
 pub const DEFAULT_DISPLAY_STATES: DisplayStates = DisplayStates {
     modified: true,
     added: true,
     removed: true,
     clean: false,
     deleted: true,
     unknown: true,
     ignored: false,
 };
 
 pub const ALL_DISPLAY_STATES: DisplayStates = DisplayStates {
     modified: true,
     added: true,
     removed: true,
     clean: true,
     deleted: true,
     unknown: true,
     ignored: true,
 };
 
 impl DisplayStates {
     pub fn is_empty(&self) -> bool {
         !(self.modified
             || self.added
             || self.removed
             || self.clean
             || self.deleted
             || self.unknown
             || self.ignored)
     }
 }
 
 fn has_unfinished_merge(repo: &Repo) -> Result<bool, CommandError> {
     return Ok(repo.dirstate_parents()?.is_merge());
 }
 
 fn has_unfinished_state(repo: &Repo) -> Result<bool, CommandError> {
     // These are all the known values for the [fname] argument of
     // [addunfinished] function in [state.py]
     let known_state_files: &[&str] = &[
         "bisect.state",
         "graftstate",
         "histedit-state",
         "rebasestate",
         "shelvedstate",
         "transplant/journal",
         "updatestate",
     ];
     if has_unfinished_merge(repo)? {
         return Ok(true);
     };
     for f in known_state_files {
         if repo.hg_vfs().join(f).exists() {
             return Ok(true);
         }
     }
     return Ok(false);
 }
 
 pub fn run(invocation: &crate::CliInvocation) -> Result<(), CommandError> {
     // TODO: lift these limitations
     if invocation
         .config
         .get(b"commands", b"status.terse")
         .is_some()
     {
         return Err(CommandError::unsupported(
             "status.terse is not yet supported with rhg status",
         ));
     }
 
     let ui = invocation.ui;
     let config = invocation.config;
     let args = invocation.subcommand_args;
 
     let verbose = !args.is_present("print0")
         && (args.is_present("verbose")
             || config.get_bool(b"ui", b"verbose")?
             || config.get_bool(b"commands", b"status.verbose")?);
 
     let all = args.is_present("all");
     let display_states = if all {
         // TODO when implementing `--quiet`: it excludes clean files
         // from `--all`
         ALL_DISPLAY_STATES
     } else {
         let requested = DisplayStates {
             modified: args.is_present("modified"),
             added: args.is_present("added"),
             removed: args.is_present("removed"),
             clean: args.is_present("clean"),
             deleted: args.is_present("deleted"),
             unknown: args.is_present("unknown"),
             ignored: args.is_present("ignored"),
         };
         if requested.is_empty() {
             DEFAULT_DISPLAY_STATES
         } else {
             requested
         }
     };
     let no_status = args.is_present("no-status");
     let list_copies = all
         || args.is_present("copies")
         || config.get_bool(b"ui", b"statuscopies")?;
 
     let repo = invocation.repo?;
 
     if verbose {
         if has_unfinished_state(repo)? {
             return Err(CommandError::unsupported(
                 "verbose status output is not supported by rhg (and is needed because we're in an unfinished operation)",
             ));
         };
     }
 
     let mut dmap = repo.dirstate_map_mut()?;
 
     let options = StatusOptions {
         // we're currently supporting file systems with exec flags only
         // anyway
         check_exec: true,
         list_clean: display_states.clean,
         list_unknown: display_states.unknown,
         list_ignored: display_states.ignored,
         list_copies,
         collect_traversed_dirs: false,
     };
 
     type StatusResult<'a> =
         Result<(DirstateStatus<'a>, Vec<PatternFileWarning>), StatusError>;
 
     let after_status = |res: StatusResult| -> Result<_, CommandError> {
         let (mut ds_status, pattern_warnings) = res?;
         for warning in pattern_warnings {
             ui.write_stderr(&print_pattern_file_warning(&warning, &repo))?;
         }
 
         for (path, error) in ds_status.bad {
             let error = match error {
                 hg::BadMatch::OsError(code) => {
                     std::io::Error::from_raw_os_error(code).to_string()
                 }
                 hg::BadMatch::BadType(ty) => {
                     format!("unsupported file type (type is {})", ty)
                 }
             };
             ui.write_stderr(&format_bytes!(
                 b"{}: {}\n",
                 path.as_bytes(),
                 error.as_bytes()
             ))?
         }
         if !ds_status.unsure.is_empty() {
             info!(
                 "Files to be rechecked by retrieval from filelog: {:?}",
                 ds_status.unsure.iter().map(|s| &s.path).collect::<Vec<_>>()
             );
         }
         let mut fixup = Vec::new();
         if !ds_status.unsure.is_empty()
             && (display_states.modified || display_states.clean)
         {
             let p1 = repo.dirstate_parents()?.p1;
             let manifest = repo.manifest_for_node(p1).map_err(|e| {
                 CommandError::from((e, &*format!("{:x}", p1.short())))
             })?;
             let working_directory_vfs = repo.working_directory_vfs();
             let store_vfs = repo.store_vfs();
             let res: Vec<_> = ds_status
                 .unsure
                 .into_par_iter()
                 .map(|to_check| {
                     unsure_is_modified(
                         working_directory_vfs,
                         store_vfs,
                         &manifest,
                         &to_check.path,
                     )
                     .map(|modified| (to_check, modified))
                 })
                 .collect::<Result<_, _>>()?;
             for (status_path, is_modified) in res.into_iter() {
                 if is_modified {
                     if display_states.modified {
                         ds_status.modified.push(status_path);
                     }
                 } else {
                     if display_states.clean {
                         ds_status.clean.push(status_path.clone());
                     }
                     fixup.push(status_path.path.into_owned())
                 }
             }
         }
         let relative_paths = config
             .get_option(b"commands", b"status.relative")?
             .unwrap_or(config.get_bool(b"ui", b"relative-paths")?);
         let output = DisplayStatusPaths {
             ui,
             no_status,
             relativize: if relative_paths {
                 Some(RelativizePaths::new(repo)?)
             } else {
                 None
             },
         };
         if display_states.modified {
             output.display(b"M ", "status.modified", ds_status.modified)?;
         }
         if display_states.added {
             output.display(b"A ", "status.added", ds_status.added)?;
         }
         if display_states.removed {
             output.display(b"R ", "status.removed", ds_status.removed)?;
         }
         if display_states.deleted {
             output.display(b"! ", "status.deleted", ds_status.deleted)?;
         }
         if display_states.unknown {
             output.display(b"? ", "status.unknown", ds_status.unknown)?;
         }
         if display_states.ignored {
             output.display(b"I ", "status.ignored", ds_status.ignored)?;
         }
         if display_states.clean {
             output.display(b"C ", "status.clean", ds_status.clean)?;
         }
 
         let dirstate_write_needed = ds_status.dirty;
         let filesystem_time_at_status_start =
             ds_status.filesystem_time_at_status_start;
 
         Ok((
             fixup,
             dirstate_write_needed,
             filesystem_time_at_status_start,
         ))
     };
     let (narrow_matcher, narrow_warnings) = narrow::matcher(repo)?;
     let (sparse_matcher, sparse_warnings) = sparse::matcher(repo)?;
     let matcher = match (repo.has_narrow(), repo.has_sparse()) {
         (true, true) => {
             Box::new(IntersectionMatcher::new(narrow_matcher, sparse_matcher))
         }
         (true, false) => narrow_matcher,
         (false, true) => sparse_matcher,
         (false, false) => Box::new(AlwaysMatcher),
     };
 
     for warning in narrow_warnings.into_iter().chain(sparse_warnings) {
         match &warning {
             sparse::SparseWarning::RootWarning { context, line } => {
                 let msg = format_bytes!(
                     b"warning: {} profile cannot use paths \"
                     starting with /, ignoring {}\n",
                     context,
                     line
                 );
                 ui.write_stderr(&msg)?;
             }
             sparse::SparseWarning::ProfileNotFound { profile, rev } => {
                 let msg = format_bytes!(
                     b"warning: sparse profile '{}' not found \"
                     in rev {} - ignoring it\n",
                     profile,
                     rev
                 );
                 ui.write_stderr(&msg)?;
             }
             sparse::SparseWarning::Pattern(e) => {
                 ui.write_stderr(&print_pattern_file_warning(e, &repo))?;
             }
         }
     }
     let (fixup, mut dirstate_write_needed, filesystem_time_at_status_start) =
         dmap.with_status(
             matcher.as_ref(),
             repo.working_directory_path().to_owned(),
             ignore_files(repo, config),
             options,
             after_status,
         )?;
 
+    // Development config option to test write races
+    if let Err(e) =
+        debug_wait_for_file(&config, "status.pre-dirstate-write-file")
+    {
+        ui.write_stderr(e.as_bytes()).ok();
+    }
+
     if (fixup.is_empty() || filesystem_time_at_status_start.is_none())
         && !dirstate_write_needed
     {
         // Nothing to update
         return Ok(());
     }
 
     // Update the dirstate on disk if we can
     let with_lock_result =
         repo.try_with_wlock_no_wait(|| -> Result<(), CommandError> {
             if let Some(mtime_boundary) = filesystem_time_at_status_start {
                 for hg_path in fixup {
                     use std::os::unix::fs::MetadataExt;
                     let fs_path = hg_path_to_path_buf(&hg_path)
                         .expect("HgPath conversion");
                     // Specifically do not reuse `fs_metadata` from
                     // `unsure_is_clean` which was needed before reading
                     // contents. Here we access metadata again after reading
                     // content, in case it changed in the meantime.
                     let fs_metadata = repo
                         .working_directory_vfs()
                         .symlink_metadata(&fs_path)?;
                     if let Some(mtime) =
                         TruncatedTimestamp::for_reliable_mtime_of(
                             &fs_metadata,
                             &mtime_boundary,
                         )
                         .when_reading_file(&fs_path)?
                     {
                         let mode = fs_metadata.mode();
                         let size = fs_metadata.len();
                         dmap.set_clean(&hg_path, mode, size as u32, mtime)?;
                         dirstate_write_needed = true
                     }
                 }
             }
             drop(dmap); // Avoid "already mutably borrowed" RefCell panics
             if dirstate_write_needed {
                 repo.write_dirstate()?
             }
             Ok(())
         });
     match with_lock_result {
         Ok(closure_result) => closure_result?,
         Err(LockError::AlreadyHeld) => {
             // Not updating the dirstate is not ideal but not critical:
             // don’t keep our caller waiting until some other Mercurial
             // process releases the lock.
             log::info!("not writing dirstate from `status`: lock is held")
         }
         Err(LockError::Other(HgError::IoError { error, .. }))
             if error.kind() == io::ErrorKind::PermissionDenied =>
         {
             // `hg status` on a read-only repository is fine
         }
         Err(LockError::Other(error)) => {
             // Report other I/O errors
             Err(error)?
         }
     }
     Ok(())
 }
 
 fn ignore_files(repo: &Repo, config: &Config) -> Vec<PathBuf> {
     let mut ignore_files = Vec::new();
     let repo_ignore = repo.working_directory_vfs().join(".hgignore");
     if repo_ignore.exists() {
         ignore_files.push(repo_ignore)
     }
     for (key, value) in config.iter_section(b"ui") {
         if key == b"ignore" || key.starts_with(b"ignore.") {
             let path = get_path_from_bytes(value);
             // TODO: expand "~/" and environment variable here, like Python
             // does with `os.path.expanduser` and `os.path.expandvars`
 
             let joined = repo.working_directory_path().join(path);
             ignore_files.push(joined);
         }
     }
     ignore_files
 }
 
 struct DisplayStatusPaths<'a> {
     ui: &'a Ui,
     no_status: bool,
     relativize: Option<RelativizePaths>,
 }
 
 impl DisplayStatusPaths<'_> {
     // Probably more elegant to use a Deref or Borrow trait rather than
     // harcode HgPathBuf, but probably not really useful at this point
     fn display(
         &self,
         status_prefix: &[u8],
         label: &'static str,
         mut paths: Vec<StatusPath<'_>>,
     ) -> Result<(), CommandError> {
         paths.sort_unstable();
         // TODO: get the stdout lock once for the whole loop
         // instead of in each write
         for StatusPath { path, copy_source } in paths {
             let relative;
             let path = if let Some(relativize) = &self.relativize {
                 relative = relativize.relativize(&path);
                 &*relative
             } else {
                 path.as_bytes()
             };
             // TODO: Add a way to use `write_bytes!` instead of `format_bytes!`
             // in order to stream to stdout instead of allocating an
             // itermediate `Vec<u8>`.
             if !self.no_status {
                 self.ui.write_stdout_labelled(status_prefix, label)?
             }
             self.ui
                 .write_stdout_labelled(&format_bytes!(b"{}\n", path), label)?;
             if let Some(source) = copy_source {
                 let label = "status.copied";
                 self.ui.write_stdout_labelled(
                     &format_bytes!(b"  {}\n", source.as_bytes()),
                     label,
                 )?
             }
         }
         Ok(())
     }
 }
 
 /// Check if a file is modified by comparing actual repo store and file system.
 ///
 /// This meant to be used for those that the dirstate cannot resolve, due
 /// to time resolution limits.
 fn unsure_is_modified(
     working_directory_vfs: hg::vfs::Vfs,
     store_vfs: hg::vfs::Vfs,
     manifest: &Manifest,
     hg_path: &HgPath,
 ) -> Result<bool, HgError> {
     let vfs = working_directory_vfs;
     let fs_path = hg_path_to_path_buf(hg_path).expect("HgPath conversion");
     let fs_metadata = vfs.symlink_metadata(&fs_path)?;
     let is_symlink = fs_metadata.file_type().is_symlink();
     // TODO: Also account for `FALLBACK_SYMLINK` and `FALLBACK_EXEC` from the
     // dirstate
     let fs_flags = if is_symlink {
         Some(b'l')
     } else if has_exec_bit(&fs_metadata) {
         Some(b'x')
     } else {
         None
     };
 
     let entry = manifest
         .find_by_path(hg_path)?
         .expect("ambgious file not in p1");
     if entry.flags != fs_flags {
         return Ok(true);
     }
     let filelog = hg::filelog::Filelog::open_vfs(&store_vfs, hg_path)?;
     let fs_len = fs_metadata.len();
     let file_node = entry.node_id()?;
     let filelog_entry = filelog.entry_for_node(file_node).map_err(|_| {
         HgError::corrupted(format!(
             "filelog missing node {:?} from manifest",
             file_node
         ))
     })?;
     if filelog_entry.file_data_len_not_equal_to(fs_len) {
         // No need to read file contents:
         // it cannot be equal if it has a different length.
         return Ok(true);
     }
 
     let p1_filelog_data = filelog_entry.data()?;
     let p1_contents = p1_filelog_data.file_data()?;
     if p1_contents.len() as u64 != fs_len {
         // No need to read file contents:
         // it cannot be equal if it has a different length.
         return Ok(true);
     }
 
     let fs_contents = if is_symlink {
         get_bytes_from_os_string(vfs.read_link(fs_path)?.into_os_string())
     } else {
         vfs.read(fs_path)?
     };
     Ok(p1_contents != &*fs_contents)
 }
 
 fn print_pattern_file_warning(
     warning: &PatternFileWarning,
     repo: &Repo,
 ) -> Vec<u8> {
     match warning {
         PatternFileWarning::InvalidSyntax(path, syntax) => format_bytes!(
             b"{}: ignoring invalid syntax '{}'\n",
             get_bytes_from_path(path),
             &*syntax
         ),
         PatternFileWarning::NoSuchFile(path) => {
             let path = if let Ok(relative) =
                 path.strip_prefix(repo.working_directory_path())
             {
                 relative
             } else {
                 &*path
             };
             format_bytes!(
                 b"skipping unreadable pattern file '{}': \
                     No such file or directory\n",
                 get_bytes_from_path(path),
             )
         }
     }
 }