On 07/11/2014 01:38 PM, Jim Jagielski wrote:
On Jul 11, 2014, at 6:36 AM, jkal...@apache.org wrote:
Author: jkaluza
Date: Fri Jul 11 10:36:15 2014
New Revision: 1609680
URL: http://svn.apache.org/r1609680
Log:
mod_proxy: add ap_proxy_define_match_worker() and use it for ProxyPassMatch
and ProxyMatch section to distinguish between normal workers and workers
with regex substitutions in the name. Implement handling of such workers
in ap_proxy_get_worker(). PR 43513
Modified:
httpd/httpd/trunk/modules/proxy/mod_proxy.c
httpd/httpd/trunk/modules/proxy/mod_proxy.h
httpd/httpd/trunk/modules/proxy/proxy_util.c
Modified: httpd/httpd/trunk/modules/proxy/mod_proxy.c
URL:
http://svn.apache.org/viewvc/httpd/httpd/trunk/modules/proxy/mod_proxy.c?rev=1609680&r1=1609679&r2=1609680&view=diff
==============================================================================
--- httpd/httpd/trunk/modules/proxy/mod_proxy.c (original)
+++ httpd/httpd/trunk/modules/proxy/mod_proxy.c Fri Jul 11 10:36:15 2014
@@ -1647,15 +1647,30 @@ static const char *
new->balancer = balancer;
}
else {
- proxy_worker *worker = ap_proxy_get_worker(cmd->temp_pool, NULL, conf,
de_socketfy(cmd->pool, r));
+ proxy_worker *worker = ap_proxy_get_worker(cmd->temp_pool, NULL, conf,
new->real);
Why do we no longer de_socketfy? Looks like this breaks UDS.
This should be ok, because new->real is already de_socketfy-ied (see
line 1601)
http://svn.apache.org/viewvc/httpd/httpd/trunk/modules/proxy/mod_proxy.c?revision=1609688&view=markup#l1601
int reuse = 0;
if (!worker) {
- const char *err = ap_proxy_define_worker(cmd->pool, &worker, NULL,
conf, r, 0);
+ const char *err;
+ if (use_regex) {
+ err = ap_proxy_define_match_worker(cmd->pool, &worker, NULL,
+ conf, r, 0);
+ }
+ else {
+ err = ap_proxy_define_worker(cmd->pool, &worker, NULL,
+ conf, r, 0);
+ }
if (err)
return apr_pstrcat(cmd->temp_pool, "ProxyPass ", err, NULL);
PROXY_COPY_CONF_PARAMS(worker, conf);
- } else {
+ }
+ else if ((use_regex != 0) ^ (worker->s->is_name_matchable)) {
Minor nit: I always disliked using bitwise xor for boolean xor :/
That's from Yann's part of the patch, but I can change it if needed.
We need an mmn bump.
Will do it in next commit.
Regards,
Jan Kaluza
