I discovered a bug in my jabberd patch that I posted on 02/24/03. I have this
patch running on a production jabberd 1.4 server, and it was crashing with
the original patch, but it seems to be much more stable now. The change is
very small (basically I changed the if statement to an 'else if' after my 'if
(sessn != NULL)'). I don't know if anyone used the patch or even cares, but I
thought just in case I'd post this updated version.
Wes Morgan
50a51
> session sessn;
53c54
< xmlnode x;
---
> xmlnode x, session_priority;
70,72c71,85
< }else if(!js_mapi_call(si, e_AUTH, p, user, NULL)){
< if(jpacket_subtype(p) == JPACKET__GET)
< { /* if it's a type="get" for auth, everybody mods it and we result and return it */
---
> }else{
> /* get existing session data (if any) */
> sessn = js_session_primary(user);
> if(sessn != NULL)
> {
> session_priority = xmlnode_get_tag(si->config,"first-session-priority");
> if(session_priority != NULL)
> {
> jutil_error(p->x, TERROR_AUTH);
> }
> log_debug(ZONE,"user %s is currently online",user->user);
> }
> else if(!js_mapi_call(si, e_AUTH, p, user, NULL)){
> if(jpacket_subtype(p) == JPACKET__GET)
> { /* if it's a type="get" for auth, everybody mods it and we result and return it */
76c89
< }else{ /* type="set" that didn't get handled used to be a problem, but now auth_plain passes on failed checks so it might be normal */
---
> }else{ /* type="set" that didn't get handled used to be a problem, but now auth_plain passes on failed checks so it might be normal */
78c91,92
< }
---
> }
> }
