httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Greg Stein <>
Subject Re: 1.3.18 is tagged and rolled, ready for testing
Date Tue, 20 Feb 2001 09:43:13 GMT
On Mon, Feb 19, 2001 at 07:35:03PM -0800, Ask Bjoern Hansen wrote:
> On Mon, 19 Feb 2001, Jim Jagielski wrote:
> > at the usual place:
> looks good on FreeBSD 4.2-STABLE.
> got a warning though:
> http_vhost.c: In function `fix_hostname':
> http_vhost.c:712: warning: passing arg 1 of `ap_unescape_url' discards qualifiers from
pointer target type

Euh... shouldn't we have that looked at? This was caused by a last-minute
checkin by Dean. Martin already pointed it out on the list, and a couple
people have mentioned the compiler problem. However, it seems like nobody is
actually talking about it. Almost as if there is a drive to call it "done"
despite the problem.

Our past few releases have had little niggly problems here and there. Every
one of them. It would be nice to get a release that can just sit out there
for six months. I have a bad feeling that this may be an indicator that
we're going to have to release another.

Can we slow down and examine the issue?

>From what I can see, it is changing r->hostname in place. r->hostname is
hopefully allocated by us, so it is (technically) safe to change. It might
be a bit safer to strdup before changing, though. Otherwise, we end up
propagating the change back into (say) the r->headers_in table. That is
certainly a bit different than what happened before. Can we be sure of the
fallout? What if somebody looked at the headers, rather than r->hostname?
For example, how about CGI and the HTTP_HOST environ variable? It might now
have an unescaped hostname. All those CGI scripts that expect it to be
unescaped will see something else.


Greg Stein,

View raw message