Re: [PATCH] issue 1954 - v6

kfogel@collab.net writes:
> I *think* this resolves issue #1954, but I'm not positive. There
> might still be places where we should be doing UTF8 checks, for
> example, since svn_path_check_valid() doesn't. However, that was no
> reason not to apply this patch -- it was complete as is. Did you
> already vet the call flow for UTF8 checks?

Aha -- right, I know what I forgot: URL operations. As Philip Martin
pointed out in

   http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgNo=86522

we have to protect 'svn mkdir URL' as well. (Perhaps this was what
your RA->check_path() changes were about? Although I think that's not
the way to go about it, if so.)

I guess that means we can't close issue #1954 quite yet. I've tweaked
my log message for r12581.

Let's figure out how to protect remote operations. Philip hinted at a
libsvn_client check, but I think a libsvn_repos (or libsvn_fs) check
would be better. Hmmm... actually, we should do *both* sides,
because that way even if the client is upgraded while the server is
not, there is still some protection.

Thoughts?

Ball in your court :-),
-Karl

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Jan 4 04:04:16 2005