On Sun, 2007-02-25 at 17:34 +0100, Rainer Jung wrote: > Two more questions: > > [EMAIL PROTECTED] schrieb: > > Modified: tomcat/connectors/trunk/jk/native/apache-1.3/mod_jk.c > > URL: > http://svn.apache.org/viewvc/tomcat/connectors/trunk/jk/native/apache-1.3/mod_jk.c?view=diff&rev=511326&r1=511325&r2=511326 > > > ============================================================================== > > --- tomcat/connectors/trunk/jk/native/apache-1.3/mod_jk.c (original) > > +++ tomcat/connectors/trunk/jk/native/apache-1.3/mod_jk.c Sat Feb 24 > 11:02:40 2007 > > @@ -1744,9 +1744,9 @@ > > jk_server_conf_t *conf = > > (jk_server_conf_t *) ap_get_module_config(s->module_config, > > &jk_module); > > - > > + > > if (jk_map_read_property(conf->worker_properties, line, 1, > conf->log) == JK_FALSE) > > - return ap_pstrcat(cmd->temp_pool, "Invalid JkWorkerProperty > ", line); > > + return ap_pstrcat(cmd->temp_pool, "Invalid JkWorkerProperty > ", line, NULL); > > > > return NULL; > > } > > @@ -2543,8 +2543,9 @@ > > ap_log_error(APLOG_MARK, APLOG_EMERG, s, > > "No worker file and no worker options in > httpd.conf " > > "use JkWorkerFile to set workers"); > > - return; > > } > > + ap_log_error(APLOG_MARK, APLOG_EMERG | APLOG_NOERRNO, 0, > NULL, "Error in reading worker properties"); > > + return !OK; > > This is inside jk_init which is void. I think this is correct w.r.t > the > apache 1.3 module structure.
Fixed. > > > > Modified: tomcat/connectors/trunk/jk/native/apache-2.0/mod_jk.c > > URL: > http://svn.apache.org/viewvc/tomcat/connectors/trunk/jk/native/apache-2.0/mod_jk.c?view=diff&rev=511326&r1=511325&r2=511326 > > > ============================================================================== > > --- tomcat/connectors/trunk/jk/native/apache-2.0/mod_jk.c (original) > > +++ tomcat/connectors/trunk/jk/native/apache-2.0/mod_jk.c Sat Feb 24 > 11:02:40 2007 > > @@ -2664,12 +2664,14 @@ > > 0, NULL, > > "No worker file and no worker options in > httpd.conf" > > "use JkWorkerFile to set workers"); > > - return; > > } > > + ap_log_error(APLOG_MARK, APLOG_EMERG | APLOG_NOERRNO, 0, > NULL, "Error in reading worker properties"); > > + return !OK; > > } > > > > if (jk_map_resolve_references(init_map, "worker.", 1, 1, > conf->log) == JK_FALSE) { > > - jk_error_exit(APLOG_MARK, APLOG_EMERG, s, pconf, "Error in > resolving configuration references"); > > + ap_log_error(APLOG_MARK, APLOG_EMERG | APLOG_NOERRNO, 0, > NULL, "Error in resolving configuration references"); > > + return !OK; > > } > > > > /* we add the URI->WORKER MAP since workers using AJP14 > > @@ -2701,7 +2704,7 @@ > > pconf)) != APR_SUCCESS) { > > ap_log_error(APLOG_MARK, APLOG_CRIT, rv, s, > > "mod_jk: could not create jk_log_lock"); > > - return HTTP_INTERNAL_SERVER_ERROR; > > + return !OK; > > } > > > > #if JK_NEED_SET_MUTEX_PERMS > > @@ -2710,7 +2713,7 @@ > > ap_log_error(APLOG_MARK, APLOG_CRIT, rv, s, > > "mod_jk: Could not set permissions on " > > "jk_log_lock; check User and Group > directives"); > > - return HTTP_INTERNAL_SERVER_ERROR; > > + return !OK; > > } > > #endif > > > > @@ -2728,7 +2731,7 @@ > > jk_server_conf_t *sconf = (jk_server_conf_t > *)ap_get_module_config(srv->module_config, > > > &jk_module); > > if (open_jklog(srv, pconf)) > > - return HTTP_INTERNAL_SERVER_ERROR; > > + return !OK; > > if (sconf) { > > if (!uri_worker_map_alloc(&(sconf->uw_map), > > > sconf->uri_to_context, sconf->log)) > > @@ -2776,7 +2779,8 @@ > > } > > } > > } > > - init_jk(pconf, conf, s); > > + if (init_jk(pconf, conf, s)) > > + return !OK; > > } > > } > > Apache 2.0/2.2 says, you can either return OK or DECLINED, everything > else is an error. So returning !OK means depending on an > implementation > detail of apache return constants (!OK could be equal to DECLINED). I > csn see, that the construct !OK is also used in apache code itself, > but > only in three modules. > > I would prefer to return another code, that's known to be different > form > OK and DECLINED. If you insist on súsing !OK, there's still one > "return > HTTP_INTERNAL_SERVER_ERROR;" left in line 2766. Ok it seems using HTTP_INTERNAL_SERVER_ERROR is a better idea. I have fixed it. Cheers Jean-Frederic > > Regards, > > Rainer --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
