Karl Fogel <kfogel@newton.ch.collab.net> writes:
> This Saturday, I was at Jim Blandy's house, and as we were playing
> around with Subversion, we noticed that "svn log" could fail silently
> under some circumstances. We traced it into svn_client_log() and saw
> that the code was just tossing an error instead of propagating it.
>
> So we sent me an email to remind me to fix it when I got home. I got
> home, looked at my commit emails, and saw that Philip Martin had
> already fixed it in rev 972 -- in fact, from the commit timestamp, it
> looks like he was writing the fix at the very same time Jim and I were
> discovering the bug. :-)
>
> Yessssssssssssssss.
Thanks, Philip! But the simultaneity isn't too surprising, when you
think about it. The bug is triggered by the failure of a name lookup
that normally wouldn't fail at all: if that name lookup usually
failed, that host name would never make it into the working directory
metadata in the first place. So the bug was only observable while
Collab's name servers were flaking out. And once you realize that the
bug's appearance is due to a (hopefully) transient condition, you get
pretty motivated to fix it while you can reproduce it.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 21 14:36:57 2006