For me it is about 10% CPU with my current system, but it also was 30-40% CPU before. This while there doesn't seem to be any activity going on (no chats).

I observed Polari for a while using top, and it never surpassed 2.7% CPU; I then recorded performance for a couple of hours using perf, and it still didn't consume any more than that (except upon starting).

Is it possible for you guys to profile Polari until it gets to undesirable CPU usage levels? Since I can't reproduce it, I personally wouldn't know where to start (other than looking for superfluous function calls, etc).

(In reply to Cody Welsh from comment #2)
> Is it possible for you guys to profile Polari until it gets to undesirable
> CPU usage levels? Since I can't reproduce it, I personally wouldn't know
> where to start (other than looking for superfluous function calls, etc).

Where to start is to compile GLib with some added some printfs to gmain.c, to print the names of the sources that are getting dispatched. You will most likely see some offending source appearing continuously. If it's not named then you can call g_source_set_name_by_id() after each use of the g_timeout_add() or g_idle_add() family of functions in Polari to fix that. If you're lucky it will be a source attached by Polari and you'll have found the problem; if you're unlucky, you'll have to look at the Telepathy client library as well and debug a bit more.

I haven't seen high CPU usage for quite a while. Nowadays it is usually 0% CPU. This around the same time as switching machines. Intel driver instead of nouveau, different CPU, etc. Not sure if it is the different system or maybe a newer Polari/glib/something. In any case, seems ok and I've checked for 1-2 months now.

dropping off the target list, then

Matthias, can we maybe close this as OBSOLETE?