[Pidgin] #1522: UPnP timeout period is ridiculously long

[Pidgin]

#1522: UPnP timeout period is ridiculously long
------------------------+---------------------------------------------------
  Reporter:  rgovostes  |       Owner:                                  
      Type:  defect     |      Status:  new                             
  Priority:  minor      |   Milestone:                                  
 Component:  libpurple  |     Version:  2.0.1                           
Resolution:             |    Keywords:  upnp timeout file transfer oscar
   Pending:  0          |  
------------------------+---------------------------------------------------
Comment (by rgovostes):

 I've been poring over libpurple's code for the last four hours, trying to
 figure out where the hell the bug is. As it turns out, half the code in
 UPnP is unused. The culprit seems to be that url_fetch_* in util.c never
 times out waiting for a response. I set up a log message in
 url_fetch_send_cb() to tell me when the data finishes being sent, then a
 second at the top of url_fetch_recv_cb() to tell me when a reply is
 received. No reply is received, thanks to my router's craptasticularity.

 My cruddy solution is to add `guint tima` to the _PurpleUtilFetchUrlData
 struct in util.h, then add this function to util.c :

 {{{
 static gboolean
 purple_util_fetch_url_timeout(gpointer data) {
         purple_debug_misc("util", "Timed out waiting for reply!");
         PurpleUtilFetchUrlData *gfud = data;

         gfud->webdata = g_realloc(gfud->webdata, gfud->len + 1);
         gfud->webdata[gfud->len] = '\0';

         gfud->callback(gfud, gfud->user_data, gfud->webdata, gfud->len,
 NULL);

         purple_util_fetch_url_cancel(gfud);
         return FALSE;
 }
 }}}

 In purple_util_fetch_url_cancel() in util.c, add

 {{{
 if (gfud->tima > 0)
                 purple_timeout_remove(gfud->tima);
 }}}

 and finally, to the bottom of url_fetch_send_cb() in util.c:

 {{{
 gfud->tima = purple_timeout_add(30000, purple_util_fetch_url_cancel,
 gfud);
 }}}

 Obviously this should be improved upon before being submitted as a patch;
 the timeout should also be in place for connecting, sending, etc. I think
 that the way I have it, there's a thirty second limit for the entire
 response to be received, which probably isn't optimal.

 With these changes, I've reduced the delay in sending a file to around one
 minute (30 seconds each for adding a port then deleting it).

-- 
Ticket URL: <http://developer.pidgin.im/ticket/1522#comment:1>
Pidgin <http://pidgin.im>
Pidgin