> From: Robert Spier [mailto:rspier@pobox.com]
> Sent: Wednesday, July 23, 2003 5:49 AM
> Sander (et al.)
>
> Apache 2.0.47 doesn't provide enough information to apr_parse_uri so
> it can parse something useful out of all uris. (see below)
>
> so.. this comparision will blow up:
>
> if (strcmp(parsed_dest_uri.hostname, r->parsed_uri.hostname)
> || strcmp(parsed_dest_uri.scheme, r->parsed_uri.scheme)) {
>
> hostname and scheme happen to be two that we don't necessarily get.
Urgh. Ok. I'll investigate this on the httpd end. IMO, these should
be filled out properly.
> so theres the following patch -- but I'm not sure I like it. It
> basically just falls through if it doesn't receive a hostname or
> scheme from apache. We could get the hostname from r->hostname, but
> this may have different meaning.
>
> Is there some deep apache juju that I'm missing?
>
> -R
>
> * mod_authz_svn.c - fix a segfault and a logic error in the MOVE/COPY
> portion of the test (r->parsed_uri was not filled out
I'll remove the check for now. That is effectively what your patch does
aswell.
> and order of arguments to strncmp is important)
Well, this is a new one for me. I can't really tell that from man strncmp.
Give me clue on this one please.
Sander
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Jul 23 11:19:06 2003