view tests/test-double-merge.t @ 40056:324b4b10351e

revlog: rewrite censoring logic

I was able to corrupt a revlog relatively easily with the existing
censoring code. The underlying problem is that the existing code
doesn't fully take delta chains into account. When copying revisions
that occur after the censored revision, the delta base can refer
to a censored revision. Then at read time, things blow up due to the
revision data not being a compressed delta.

This commit rewrites the revlog censoring code to take a higher-level
approach. We now create a new revlog instance pointing at temp files.
We iterate through each revision in the source revlog and insert
those revisions into the new revlog, replacing the censored revision's
data along the way.

The new implementation isn't as efficient as the old one. This is
because it will fully engage delta computation on insertion. But I
don't think it matters.

The new implementation is a bit hacky because it attempts to reload
the revlog instance with a new revlog index/data file. This is fragile.
But this is needed because the index (which could be backed by C) would
have a cached copy of the old, possibly changed data and that could
lead to problems accessing index or revision data later.

One benefit of the new approach is that we integrate with the
transaction. The old revlog is backed up and if the transaction is
rolled back, the original revlog is restored.

As part of this, we had to teach the transaction about the store
vfs. I'm not super keen about this. But this was the easiest way
to hook things up to the transaction. We /could/ just ignore the
transaction like we were doing before. But any file mutation should
be governed by transaction semantics, including undo during rollback.

Differential Revision: https://phab.mercurial-scm.org/D4869

  $ hg init repo
  $ cd repo

  $ echo line 1 > foo
  $ hg ci -qAm 'add foo'

copy foo to bar and change both files
  $ hg cp foo bar
  $ echo line 2-1 >> foo
  $ echo line 2-2 >> bar
  $ hg ci -m 'cp foo bar; change both'

in another branch, change foo in a way that doesn't conflict with
the other changes
  $ hg up -qC 0
  $ echo line 0 > foo
  $ hg cat foo >> foo
  $ hg ci -m 'change foo'
  created new head

we get conflicts that shouldn't be there
  $ hg merge -P
  changeset:   1:484bf6903104
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     cp foo bar; change both
  
  $ hg merge --debug
    searching for copies back to rev 1
    unmatched files in other:
     bar
    all copies found (* = to merge, ! = divergent, % = renamed and deleted):
     src: 'foo' -> dst: 'bar' *
    checking for directory renames
  resolving manifests
   branchmerge: True, force: False, partial: False
   ancestor: e6dc8efe11cc, local: 6a0df1dad128+, remote: 484bf6903104
   preserving foo for resolve of bar
   preserving foo for resolve of foo
  starting 4 threads for background file closing (?)
   bar: remote copied from foo -> m (premerge)
  picked tool ':merge' for bar (binary False symlink False changedelete False)
  merging foo and bar to bar
  my bar@6a0df1dad128+ other bar@484bf6903104 ancestor foo@e6dc8efe11cc
   premerge successful
   foo: versions differ -> m (premerge)
  picked tool ':merge' for foo (binary False symlink False changedelete False)
  merging foo
  my foo@6a0df1dad128+ other foo@484bf6903104 ancestor foo@e6dc8efe11cc
   premerge successful
  0 files updated, 2 files merged, 0 files removed, 0 files unresolved
  (branch merge, don't forget to commit)

contents of foo
  $ cat foo
  line 0
  line 1
  line 2-1

contents of bar
  $ cat bar
  line 0
  line 1
  line 2-2

  $ cd ..