On 10/12/2005 12:28 AM, Jim Jagielski wrote: > Ruediger Pluem wrote:
[..cut..] >>BTW: I just noticed that >> >> c = strchr(url, ':'); >> if (c == NULL || c[1] != '/' || c[2] != '/' || c[3] == '\0') >> return NULL; >> >>in ap_proxy_get_worker of proxy_util.c seem to be unneeded now >>(Only came to my mind as I saw your commit mail, weird. ) >>If you agree I will remove them tomorrow as a clean-up. >> > > > It's a quick way, I think, I avoiding bogus urls (one's that > lack schemes) without wasting time with the various > strncasecmps. Yes, this is true, but how often will this quick way be used in a correctly configured system? > > Of course, that get's me thinking... maybe strncmp() is more > correct. We want to be case ignorant with hostname but NOT > with path. And, iirc, we force the hostname in name to be > lowercase. Let me look into that :) > Very good point. From what I see in ap_proxy_add_worker we force the hostname and only the hostname to be lowercase by parsing / unparsing the URL, BUT: 1. We do not do so currently with the right side of ProxyPass (real name). 2. We not do this in proxy_handler with the URL we pass to to ap_proxy_pre_request. I think it would belong here to ensure that we also handle requests from mod_rewrite correctly. Maybe we need to do the whole URL normalization by parsing the URL, lowercasing the hostname and unparsing the URL also in ap_proxy_get_worker. 3. The whole URL is converted to lowercase for BalancerMembers in line 1438 of mod_proxy.c before this name is added as a worker :-(. So some work is left over to fix this. Curious to see what additional things you will find :). Regards Rüdiger
