Re: nasty commit bug

Karl Fogel <kfogel@newton.ch.collab.net> writes:
> Okay -- and this is again a match with Ben's hypothesis, as though
> your text bases aren't being updated, even while your entries files
> are.

By the way, a thought:

I can't remember whether we write a new entry or the corresponding
text base first, but it's not supposed to matter -- because the new
text base should definitely be ready in tmp/text-base/ before the new
entry is written, and the working copy dir should be locked until
everything succeeds.

However, if something were to seg fault (ahem), such that the entry
got updated before the tmp/text-base were moved, AND for some reason
the dir also got unlocked when it shouldn't have been (and/or the
.svn/log file removed because erroneously thought done), then we might
see this bug.

Just a tentative hypothesis, may be totally wrong. Going to bed now;
good luck to people in other time zones, and people who keep true
hackers' hours :-).

-K

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Jul 13 08:07:07 2002