cmpilato@collab.net wrote:
>Ben Collins-Sussman <sussman@collab.net> writes:
>
>
>
>>>+ When a commit fails for any reason, the transaction it created in the
>>>+ repository is not removed. This means that the repository
>>>+ administrator must use svnadmin to remove stale transactions periodically.
>>>+
>>>+ The XFAIL test for this issue is commit_tests.py 24.
>>>
>>>
>>Why do you make such a sweeping statement, Brane?
>>
>>I'm surprised, because we call txn_abort() all over the place in our
>>code, if any kind of svn_error_t is thrown at any point during a
>>commit. This is a really old feature that we wrote in the beginning,
>>and tested a lot. If we've regressed, I'm amazed. I'll check out
>>your new xfail test.
>>
>>
>
>Heh -- wasn't it just a few days ago that you and I were rescuing lost
>log messages from their non-aborted txns when people's commits failed
>(gstein's in my case, and then someone else you assisted) ? So, you
>shouldn't be *too* surprised.
>
>Brane, was that blanket statement accurate? Is this a
>commits-over-DAV only thing, or ... help us understand the scope of
>the problem, please.
>
>
It's both DAV and local, and I verified it with both server-induced
(conflict) and client-induced (eoln translation) commit failures. Sure,
that isn't "any" reason, but it covers quite a lot, IMHO.
--
Brane Čibej <brane_at_xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Sep 27 16:56:55 2002