Re: [Rhythmbox-devel] RBShellPlayer and MonkeyMediaPlayer finalizationbug

["Benjamin Liblit" <liblit eecs berkeley edu>]

Benjamin Otte wrote:
> Considering you can just use [...] and be done with it - without 
> getting that bug again when someone decides to ref the player at some
> other place - that sounds a lot better to me.

Your robustness argument is compelling, Benjamin.  Furthermore, I have 
looked into the origins of those additional refs in greater detail, and 
they would be quite difficult to prevent.  Many of them arise, for 
example, from timeout-delayed callbacks within MonkeyMediaPlayer itself.

OK, removing the callbacks via g_signal_handlers_disconnect_matched() 
seems to be right thing.  There are some eel_gconf_* notification 
callbacks that ought to be removed in the finalization method as well. 
I've already tested a fix that does all of this, and it looks good. 
I'll post a Bugzilla bug report with description + patch shortly.

Thanks for the feedback, Benjamin!