[PATCH 2 of 2 STABLE] _adjustlinkrev: reuse ancestors set during rename detection (issue4514)

[Martin von Zweigbergk]

To the degree I understand the code, this looks "good" to me (under the
circumstances). Thanks!

On Fri Jan 30 2015 at 9:22:36 AM Pierre-Yves David <
pierre-yves.david at ens-lyon.org> wrote:

> # HG changeset patch
> # User Pierre-Yves David <pierre-yves.david at fb.com>
> # Date 1422633748 0
> #      Fri Jan 30 16:02:28 2015 +0000
> # Branch stable
> # Node ID 9e07d1f5d85ad76fa4a6b0191a54b9bcdac1442c
> # Parent  b9d18d103d7aa91ac999630ec582300a5cb9a596
> _adjustlinkrev: reuse ancestors set during rename detection (issue4514)
>
> The new linkrev adjustement mechanism makes renames detection very slow,
> because
> each file rewalk the ancestors dag. To mitigate the issue in Mercurial
> 3.3, we
> introduce a simplistic was to share the ancestors computation for the
> linkrev
> validation phase.
>
> We can reuse the ancestors in that case because we do not care about
> sub-branching in the ancestors graph.
>
> The cached set will be use to check if the linkrev is valid in the search
> context. This is the vast majority of the ancestors usage during copies
> search
> since the uncached one will only be used when linkrev is invalid, which is
> hopefully rare.
>
> diff --git a/mercurial/context.py b/mercurial/context.py
> --- a/mercurial/context.py
> +++ b/mercurial/context.py
> @@ -764,14 +764,21 @@ class basefilectx(object):
>          cl = repo.unfiltered().changelog
>          ma = repo.manifest
>          # fetch the linkrev
>          fr = filelog.rev(fnode)
>          lkr = filelog.linkrev(fr)
> +        # hack to reuse ancestor computation when searching for renames
> +        memberanc = getattr(self, '_ancestrycontext', None)
> +        iteranc = None
> +        if memberanc is None:
> +            memberanc = iteranc = cl.ancestors([srcrev], lkr,
> +                                               inclusive=inclusive)
>          # check if this linkrev is an ancestor of srcrev
> -        anc = cl.ancestors([srcrev], lkr, inclusive=inclusive)
> -        if lkr not in anc:
> -            for a in anc:
> +        if lkr not in memberanc:
> +            if iteranc is None:
> +                iteranc = cl.ancestors([srcrev], lkr, inclusive=inclusive)
> +            for a in iteranc:
>                  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 == ma.readfast(ac[0]).get(path):
> @@ -824,10 +831,12 @@ class basefilectx(object):
>                  # fed), ensure the created filectx is associated with a
>                  # changeset that is an ancestor of self.changectx.
>                  rev = self._adjustlinkrev(path, l, fnode, self.rev())
>                  fctx = filectx(self._repo, path, fileid=fnode, filelog=l,
>                                 changeid=rev)
> +                fctx._ancestrycontext = getattr(self, '_ancestrycontext',
> None)
> +
>              else:
>                  fctx = filectx(self._repo, path, fileid=fnode, filelog=l)
>              ret.append(fctx)
>          return ret
>
> diff --git a/mercurial/copies.py b/mercurial/copies.py
> --- a/mercurial/copies.py
> +++ b/mercurial/copies.py
> @@ -129,10 +129,11 @@ def _chain(src, dst, a, b):
>  def _tracefile(fctx, am, limit=-1):
>      '''return file context that is the ancestor of fctx present in
> ancestor
>      manifest am, stopping after the first ancestor lower than limit'''
>
>      for f in fctx.ancestors():
> +        assert f._ancestrycontext is not None
>          if am.get(f.path(), None) == f.filenode():
>              return f
>          if f.rev() < limit:
>              return None
>
> @@ -168,12 +169,15 @@ def _forwardcopies(a, b):
>      # this means we can miss a case like 'hg rm b; hg cp a b'
>      cm = {}
>      missing = set(b.manifest().iterkeys())
>      missing.difference_update(a.manifest().iterkeys())
>
> +    ancestrycontext = a._repo.changelog.ancestors([b.rev()],
> inclusive=True)
>      for f in missing:
> -        ofctx = _tracefile(b[f], am, limit)
> +        fctx = b[f]
> +        fctx._ancestrycontext = ancestrycontext
> +        ofctx = _tracefile(fctx, am, limit)
>          if ofctx:
>              cm[f] = ofctx.path()
>
>      # combine copies from dirstate if necessary
>      if w is not None:
> _______________________________________________
> Mercurial-devel mailing list
> Mercurial-devel at selenic.com
> http://selenic.com/mailman/listinfo/mercurial-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://selenic.com/pipermail/mercurial-devel/attachments/20150130/f1fddcbd/attachment.html>