[RFC][merge] Patience Sorting Diffs

[John Arbash Meinel]

John Arbash Meinel wrote:

...

> I did run a test which upgraded an old v4 bzr.dev archive to the new
> weave format, using cdv diff instead of difflib. And 'bzr check' works.
> I'm pretty positive that I generate valid diffs (since I mostly just use
> difflibs builtin code to generate them). They may not be optimal, but I
> think they are better than difflib.
> 
> I would like to see this code merged in sometime, but I would first like
> people to play with cdvdiff and see what they think.
> (The command to run is ./bzrlib/cdv/cdvdifflib.py file_a file_b, though
> the builtin 'bzr diff' has also been updated).

I have to back off this comment for a little bit. I my cdv-diff branch
passes all tests, but when I try to do a "bzr upgrade" I get an
assertion error. Something about the revision being deleted not existing
in the active entries.

else:
  assert c == ']'
  if v in included:
    assert v in dset #This assert is failing
    dset.remove(v)

I'll check into it more thoroughly, and hopefully I'll find the problem.
(My best guess is that cdvdifflib.SequenceMatcher is returning blocks
differently than what is expected, so the alignment of {} and [] is not
the same. However, in my direct tests of this, it returns the expected
blocks...)


> 
> John
> =:->

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 256 bytes
Desc: OpenPGP digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20051212/d42485b1/attachment.pgp