Philip Martin <philip@codematters.co.uk> writes:
> sussman@tigris.org writes:
>
> > Author: sussman
> > Date: 2002-09-05 16:11:38 -0500 (Thu, 05 Sep 2002)
> > New Revision: 3138
> >
> > Modified:
> > trunk/subversion/libsvn_ra_dav/props.c
> > Log:
> >
> > Heh, ra_dav did a PROPFIND on /A/B, which failed, then on /A, which
> > failed, then on /, which failed. But then it got stuck in an infinite
> > loop, because svn_path_remove_component("/") == "/".
>
> Hmm, I changed svn_path_remove_component in rev 3113, but I don't
> think I changed the behaviour of svn_path_remove_component("/"). I
> also fixed an infinite loop bug in that revision, but I only triggered
> that when I had an error in my code. Now you have found another such
> bug, and a quick grep through the code reveals suspicious loops in
> okay_to_bump_path in ra_dav/merge.c and svn_ra_local__split_URL in
> ra_local/split_url.c. Do those terminate? Did I make a mistake in
> rev 3113? Should svn_path_remove_component("/") return "/"?
I must have not had r3113 when I hit this infinite loop. I think
there's nothing wrong with svn_path_remove_component("/") returning
"/".
Maybe we should check out those other infinite loop situations, yes.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Sep 6 00:11:40 2002