# HG changeset patch
# User Valentin Gatien-Baron <valentin.gatienbaron@gmail.com>
# Date 2019-07-14 03:45:32
# Node ID d98ec36be80852b864779af7c4b77797fdeb9b76
# Parent  20d0e59be79b0c27d76b8913949ec076ef81a18d

convert: add a config option to help doing identity hg->hg conversion

I want to change the computation of the list of files modified by a
commit. In principle, this would simply change a cache. But since this
information is stored in commits rather than a cache, changing it
means changing commit hashes (going forward).

Some users rely on the convert extension from hg to hg not changing
hashes when nothing changes (usually). Allow these users to preserve
hashes despite changes to the changelog files computation by reusing
these files lists when the manifest is unchanged (since these files
list are derived from the manifest).

Differential Revision: https://phab.mercurial-scm.org/D6643

diff --git a/hgext/convert/__init__.py b/hgext/convert/__init__.py
--- a/hgext/convert/__init__.py
+++ b/hgext/convert/__init__.py
@@ -439,6 +439,11 @@ def convert(ui, src, dest=None, revmapfi
     :convert.hg.sourcename: records the given string as a 'convert_source' extra
         value on each commit made in the target repository. The default is None.
 
+    :convert.hg.preserve-hash: only works with mercurial sources. Make convert
+        prevent performance improvement to the list of modified files in commits
+        when such an improvement would cause the hash of a commit to change.
+        The default is False.
+
     All Destinations
     ################
 
diff --git a/hgext/convert/common.py b/hgext/convert/common.py
--- a/hgext/convert/common.py
+++ b/hgext/convert/common.py
@@ -114,7 +114,7 @@ SKIPREV = 'SKIP'
 class commit(object):
     def __init__(self, author, date, desc, parents, branch=None, rev=None,
                  extra=None, sortkey=None, saverev=True, phase=phases.draft,
-                 optparents=None):
+                 optparents=None, ctx=None):
         self.author = author or 'unknown'
         self.date = date or '0 0'
         self.desc = desc
@@ -126,6 +126,7 @@ class commit(object):
         self.sortkey = sortkey
         self.saverev = saverev
         self.phase = phase
+        self.ctx = ctx # for hg to hg conversions
 
 class converter_source(object):
     """Conversion source interface"""
diff --git a/hgext/convert/hg.py b/hgext/convert/hg.py
--- a/hgext/convert/hg.py
+++ b/hgext/convert/hg.py
@@ -339,7 +339,11 @@ class mercurial_sink(common.converter_si
                                    phases.phasenames[commit.phase], 'convert')
 
             with self.repo.transaction("convert") as tr:
-                node = nodemod.hex(self.repo.commitctx(ctx))
+                if self.repo.ui.config('convert', 'hg.preserve-hash'):
+                    origctx = commit.ctx
+                else:
+                    origctx = None
+                node = nodemod.hex(self.repo.commitctx(ctx, origctx=origctx))
 
                 # If the node value has changed, but the phase is lower than
                 # draft, set it back to draft since it hasn't been exposed
@@ -591,7 +595,8 @@ class mercurial_source(common.converter_
                              extra=ctx.extra(),
                              sortkey=ctx.rev(),
                              saverev=self.saverev,
-                             phase=ctx.phase())
+                             phase=ctx.phase(),
+                             ctx=ctx)
 
     def numcommits(self):
         return len(self.repo)
diff --git a/hgext/eol.py b/hgext/eol.py
--- a/hgext/eol.py
+++ b/hgext/eol.py
@@ -400,7 +400,7 @@ def reposetup(ui, repo):
                     if wlock is not None:
                         wlock.release()
 
-        def commitctx(self, ctx, error=False):
+        def commitctx(self, ctx, error=False, origctx=None):
             for f in sorted(ctx.added() + ctx.modified()):
                 if not self._eolmatch(f):
                     continue
@@ -416,6 +416,6 @@ def reposetup(ui, repo):
                 if inconsistenteol(data):
                     raise errormod.Abort(_("inconsistent newline style "
                                            "in %s\n") % f)
-            return super(eolrepo, self).commitctx(ctx, error)
+            return super(eolrepo, self).commitctx(ctx, error, origctx)
     repo.__class__ = eolrepo
     repo._hgcleardirstate()
diff --git a/hgext/keyword.py b/hgext/keyword.py
--- a/hgext/keyword.py
+++ b/hgext/keyword.py
@@ -785,8 +785,8 @@ def reposetup(ui, repo):
             finally:
                 del self.commitctx
 
-        def kwcommitctx(self, ctx, error=False):
-            n = super(kwrepo, self).commitctx(ctx, error)
+        def kwcommitctx(self, ctx, error=False, origctx=None):
+            n = super(kwrepo, self).commitctx(ctx, error, origctx)
             # no lock needed, only called from repo.commit() which already locks
             if not kwt.postcommit:
                 restrict = kwt.restrict
diff --git a/hgext/lfs/__init__.py b/hgext/lfs/__init__.py
--- a/hgext/lfs/__init__.py
+++ b/hgext/lfs/__init__.py
@@ -227,9 +227,9 @@ def _reposetup(ui, repo):
 
     class lfsrepo(repo.__class__):
         @localrepo.unfilteredmethod
-        def commitctx(self, ctx, error=False):
+        def commitctx(self, ctx, error=False, origctx=None):
             repo.svfs.options['lfstrack'] = _trackedmatcher(self)
-            return super(lfsrepo, self).commitctx(ctx, error)
+            return super(lfsrepo, self).commitctx(ctx, error, origctx=origctx)
 
     repo.__class__ = lfsrepo
 
diff --git a/hgext/remotefilelog/shallowrepo.py b/hgext/remotefilelog/shallowrepo.py
--- a/hgext/remotefilelog/shallowrepo.py
+++ b/hgext/remotefilelog/shallowrepo.py
@@ -161,7 +161,7 @@ def wraprepo(repo):
                                                               **kwargs)
 
         @localrepo.unfilteredmethod
-        def commitctx(self, ctx, error=False):
+        def commitctx(self, ctx, error=False, origctx=None):
             """Add a new revision to current repository.
             Revision information is passed via the context argument.
             """
@@ -179,7 +179,8 @@ def wraprepo(repo):
                         files.append((f, hex(fparent1)))
                 self.fileservice.prefetch(files)
             return super(shallowrepository, self).commitctx(ctx,
-                                                            error=error)
+                                                            error=error,
+                                                            origctx=origctx)
 
         def backgroundprefetch(self, revs, base=None, repack=False, pats=None,
                                opts=None):
diff --git a/mercurial/configitems.py b/mercurial/configitems.py
--- a/mercurial/configitems.py
+++ b/mercurial/configitems.py
@@ -291,6 +291,9 @@ coreconfigitem('convert', 'hg.clonebranc
 coreconfigitem('convert', 'hg.ignoreerrors',
     default=False,
 )
+coreconfigitem('convert', 'hg.preserve-hash',
+    default=False,
+)
 coreconfigitem('convert', 'hg.revs',
     default=None,
 )
diff --git a/mercurial/localrepo.py b/mercurial/localrepo.py
--- a/mercurial/localrepo.py
+++ b/mercurial/localrepo.py
@@ -2578,7 +2578,7 @@ class localrepository(object):
         return ret
 
     @unfilteredmethod
-    def commitctx(self, ctx, error=False):
+    def commitctx(self, ctx, error=False, origctx=None):
         """Add a new revision to current repository.
         Revision information is passed via the context argument.
 
@@ -2586,6 +2586,12 @@ class localrepository(object):
         modified/added/removed files. On merge, it may be wider than the
         ctx.files() to be committed, since any file nodes derived directly
         from p1 or p2 are excluded from the committed ctx.files().
+
+        origctx is for convert to work around the problem that bug
+        fixes to the files list in changesets change hashes. For
+        convert to be the identity, it can pass an origctx and this
+        function will use the same files list when it makes sense to
+        do so.
         """
 
         p1, p2 = ctx.p1(), ctx.p2()
@@ -2701,6 +2707,9 @@ class localrepository(object):
                 filesadded = filesadded or None
                 filesremoved = filesremoved or None
 
+            if origctx and origctx.manifestnode() == mn:
+                files = origctx.files()
+
             # update changelog
             self.ui.note(_("committing changelog\n"))
             self.changelog.delayupdate(tr)
diff --git a/mercurial/repository.py b/mercurial/repository.py
--- a/mercurial/repository.py
+++ b/mercurial/repository.py
@@ -1656,7 +1656,7 @@ class ilocalrepositorymain(interfaceutil
                editor=False, extra=None):
         """Add a new revision to the repository."""
 
-    def commitctx(ctx, error=False):
+    def commitctx(ctx, error=False, origctx=None):
         """Commit a commitctx instance to the repository."""
 
     def destroying():
diff --git a/tests/test-commandserver.t b/tests/test-commandserver.t
--- a/tests/test-commandserver.t
+++ b/tests/test-commandserver.t
@@ -917,13 +917,13 @@ cases.
   >     raise error.Abort(b'fail after finalization')
   > def reposetup(ui, repo):
   >     class failrepo(repo.__class__):
-  >         def commitctx(self, ctx, error=False):
+  >         def commitctx(self, ctx, error=False, origctx=None):
   >             if self.ui.configbool(b'failafterfinalize', b'fail'):
   >                 # 'sorted()' by ASCII code on category names causes
   >                 # invoking 'fail' after finalization of changelog
   >                 # using "'cl-%i' % id(self)" as category name
   >                 self.currenttransaction().addfinalize(b'zzzzzzzz', fail)
-  >             return super(failrepo, self).commitctx(ctx, error)
+  >             return super(failrepo, self).commitctx(ctx, error, origctx)
   >     repo.__class__ = failrepo
   > EOF
 
diff --git a/tests/test-convert-identity.t b/tests/test-convert-identity.t
new file mode 100644
--- /dev/null
+++ b/tests/test-convert-identity.t
@@ -0,0 +1,40 @@
+Testing that convert.hg.preserve-hash=true can be used to make hg
+convert from hg repo to hg repo preserve hashes, even if the
+computation of the files list in commits change slightly between hg
+versions.
+
+  $ cat <<'EOF' >> "$HGRCPATH"
+  > [extensions]
+  > convert =
+  > EOF
+  $ cat <<'EOF' > changefileslist.py
+  > from mercurial import (changelog, extensions)
+  > def wrap(orig, clog, manifest, files, *args, **kwargs):
+  >   return orig(clog, manifest, ["a"], *args, **kwargs)
+  > def extsetup(ui):
+  >   extensions.wrapfunction(changelog.changelog, 'add', wrap)
+  > EOF
+
+  $ hg init repo
+  $ cd repo
+  $ echo a > a; hg commit -qAm a
+  $ echo b > a; hg commit -qAm b
+  $ hg up -qr 0; echo c > c; hg commit -qAm c
+  $ hg merge -qr 1
+  $ hg commit -m_ --config extensions.x=../changefileslist.py
+  $ hg log -r . -T '{node|short} {files|json}\n'
+  c085bbe93d59 ["a"]
+
+Now that we have a commit with a files list that's not what the
+current hg version would create, check that convert either fixes it or
+keeps it depending on config:
+
+  $ hg convert -q . ../convert
+  $ hg --cwd ../convert log -r tip -T '{node|short} {files|json}\n'
+  b7c4d4bbacd3 []
+  $ rm -rf ../convert
+
+  $ hg convert -q . ../convert --config convert.hg.preserve-hash=true
+  $ hg --cwd ../convert log -r tip -T '{node|short} {files|json}\n'
+  c085bbe93d59 ["a"]
+  $ rm -rf ../convert
diff --git a/tests/test-convert.t b/tests/test-convert.t
--- a/tests/test-convert.t
+++ b/tests/test-convert.t
@@ -373,6 +373,11 @@
                     records the given string as a 'convert_source' extra value
                     on each commit made in the target repository. The default is
                     None.
+      convert.hg.preserve-hash
+                    only works with mercurial sources. Make convert prevent
+                    performance improvement to the list of modified files in
+                    commits when such an improvement would cause the hash of a
+                    commit to change. The default is False.
   
       All Destinations
       ################