On Tue, Nov 26, 2013 at 12:24 PM, Christophe Fergeau <cfergeau@xxxxxxxxxx> wrote: > On Mon, Nov 25, 2013 at 08:41:27AM -0500, Marc-André Lureau wrote: >> >> >> ----- Original Message ----- >> > On Mon, Nov 25, 2013 at 02:31:10PM +0100, Marc-André Lureau wrote: >> > > On Mon, Nov 25, 2013 at 1:41 PM, Christophe Fergeau <cfergeau@xxxxxxxxxx> >> > > wrote: >> > > > On Mon, Nov 25, 2013 at 01:33:52PM +0100, Marc-André Lureau wrote: >> > > >> Check that there is an actual stream before calling cork(). >> > > > >> > > > Is it expected that we are getting called with a non-NULL stream in the >> > > > first place? >> > > >> > > Since this is sent by server on mjpeg stream creation, I would say >> > > playback stream isn't mandatory. >> > > >> > > Anyway, it should not crash. >> > >> > Of course not, but we should fix things in the right place, hence the >> > question to know if this is the right place for that fix. >> > Your answer seems to be about playback_min_latency_changed(), what about >> > stream_uncork? Is this happening in the same situation? >> > >> >> No, there are checks for stream presence (it was checked everywhere, only >> since efa65b49f it didn't check for min-latency property changes) > > This belongs in the commit log imo, and I did not understand what the > g_return_if_fail is doing in that commit from your answer. Avoiding a call that would lead to an abort(), catching the error earlier with a warning. -- Marc-André Lureau _______________________________________________ Spice-devel mailing list Spice-devel@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/spice-devel