Core Dev opinions on Bug 395556 ?

[Martin Pool]

2009/7/8 Martin (gzlist) <gzlist at googlemail.com>:
> Okay, I am now back, though lacking my main machine again for the moment.
>
> On 07/07/2009, Martin Pool <mbp at sourcefrog.net> wrote:
>>
>> Thanks to both of you for the simple script to reproduce it.  I agree
>> it's a high priority bug.  It's surprising that it's occurring without
>> a selective commit.  It's a bit related to some rework of commit that
>> Ian and Robert have been looking at.
>
> Is it worth committing, or at least reviewing, the bundle I attached
> to the original mail? The more detailed tests included prevent some of
> the confusion Maritza had when looking just at the
> quick-steps-to-reproduce version.
>
> https://lists.ubuntu.com/archives/bazaar/attachments/20090702/63b634f3/attachment.obj
>
> Not sure having a known failure in tree is an incentive or
> disincentive to get a fix landed.

It's a step forward, and certainly can be merged.  Thanks for writing
it.  The known failure should include the url of the bug and then the
bug should mention the name of the test, so that someone who either
wants to fix the failure or continue work on the bug can get between
them.

The patch looks very clear, except that the one thing I find strange
is that it expects the commit to pass when run directly through
tree.commit(), but to fail when it's run from cmd_commit.  Is that
really true?  If so it's very interesting information towards fixing
the bug.

One more thing - if you have a patch, even if it's not a complete fix
- please put it in either a launchpad merge proposal or bundle buggy,
so that we can track it
<https://code.edge.launchpad.net/bzr/+activereviews> and be sure to
come to a specific conclusion about it.

-- 
Martin <http://launchpad.net/~mbp/>