Requesting suggestions for debugging bizarre bzr-svn merge issue

[Max Bowsher]

On 20/07/10 23:08, Jelmer Vernooij wrote:
> Hi Max,
> 
> On Wed, 07 Jul 2010 00:14:39 +0100
> Max Bowsher <maxb at f2s.com> wrote:
>> I've discovered a rather perplexing problem with bzr-svn.
>>
>> I attempted to branch and then cherrypick some changes from a bzr-svn
>> imported branch. The cherrypick merge produced some very dodgy
>> results, looking as if some lines were missing in the merge results,
>> or the conflict markers were in the wrong place.
>>
>> On a hunch, I bzr fast-exported and fast-imported the branch, and
>> redid the exact same merges. The merge results were then much saner.
>>
>> As a result I conjecture that bzr-svn is doing something broken in the
>> way it drives bzrlib when importing a branch, which gets ironed out
>> when roundtripping through a fastimport stream.
>>
>> Can anyone give me advice on what to look into?
>>
>> Regrettably the branches in question are ones I am not at liberty to
>> disclose.

> This is rather vague - is it possible to construct a testcase that has
> the same or similar issues ? 
> 
> What version of bzr-svn? Is "bzr check" on the bzr-svn-created branch
> happy?


Although I was absolutely sure that's what happened, I have since failed
to be able to reproduce it on demand - both bzr-svn and
fastimport-round-tripped branches display the same broken merge behaviour.

So, I must apologise for maligning bzr-svn, and turn my attention to the
bzr merge algorithms, and attempting to construct a testcase.

Max.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: OpenPGP digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20100721/976eb120/attachment.pgp