Well, it's been eight days since we released Pidgin 2.5.3. In that 8 days, we've had more duplicate tickets than we care to count over the all-too-common hang on exit on Windows systems. In the faint hope that people actually read my ramblings, I'm posting this here to give a brief synopsis of the problem and to discourage further duplicate tickets.
On Windows sytems, Pidgin uses threads for a few things, namely DNS lookups and Network Location Awareness (NLA) stuff. For the Linux-inclined, NLA is somewhat similar to NetworkManager. When we released Pidgin 2.5.3, we started getting a bunch of reports about hangs on exit (including a number of people who don't know the difference between a hang and a crash, but that's another post in itself). All the debug logs pointed to wpurple_cleanup(), a function we call on close to tie up some loose ends, or "clean up." One of the areas this function cleans up is NLA-related stuff.
This code hasn't changed meaningfully in quite some time, but magically it became a problem for users of 2.5.3. It doesn't really make any sense to me why this would suddenly stop working in the current release, but the previous release seems to be almost entirely problem-free in this regard. Confusion aside, however, we have a proposed fix that will be reviewed and possibly tweaked.
So in summary, we know what the problem is and we are working to fix it. Please, please, please don't open anymore duplicate tickets about it!
On Windows sytems, Pidgin uses threads for a few things, namely DNS lookups and Network Location Awareness (NLA) stuff. For the Linux-inclined, NLA is somewhat similar to NetworkManager. When we released Pidgin 2.5.3, we started getting a bunch of reports about hangs on exit (including a number of people who don't know the difference between a hang and a crash, but that's another post in itself). All the debug logs pointed to wpurple_cleanup(), a function we call on close to tie up some loose ends, or "clean up." One of the areas this function cleans up is NLA-related stuff.
This code hasn't changed meaningfully in quite some time, but magically it became a problem for users of 2.5.3. It doesn't really make any sense to me why this would suddenly stop working in the current release, but the previous release seems to be almost entirely problem-free in this regard. Confusion aside, however, we have a proposed fix that will be reviewed and possibly tweaked.
So in summary, we know what the problem is and we are working to fix it. Please, please, please don't open anymore duplicate tickets about it!
