Description of problem: In default GNOME Shell, starting Claws Mail doesn't display any window. It starts minimized. That is because during application start, Claws Mail calls gtk_window_iconify() and gtk_window_deiconify() some time later. The latter doesn't have an affect. Version-Release number of selected component (if applicable): $ rpm -q mutter mutter-2.91.93-1.fc15.x86_64
*ping* mutter maintainers: Is this misassigned? Can you please comment on this issue?
(In reply to comment #1) > *ping* mutter maintainers: Is this misassigned? Can you please comment on this > issue? I can't reproduce this with this test case: var Gtk = imports.gi.Gtk; Gtk.init(null); var w = new Gtk.Window({type: Gtk.WindowType.TOPLEVEL }); w.set_default_size(640, 480); w.add(new Gtk.Button({label: "hi" })); w.show_all(); w.iconify(); w.deiconify(); Gtk.main(); In other words calling iconify() then deiconify() appears to work.
Would need a more minmal test case to debug this then.
Thank you for replying here! A *less* (!) minimal test-case is what I've been trying to find meanwhile. The corresponding C code for your test-case cannot be used to reproduce it either. Not even with an inserted sleep(). Claws Mail sets up a lot of GTK stuff between the iconify() and deiconify() calls. It could be that Claws Mail does something bad, perhaps in its GtkCMCTree implementation, either confusing GTK or corrupting something. The following patch is the most minimal patch to Claws Mail that does not move the iconify/deiconify calls next to eachother: http://www.thewildbeast.co.uk/claws-mail/bugzilla/attachment.cgi?id=990&action=diff Based on that, mission objective is to find out whether gtk_widget_set_style(ctree, normal_style) does something bad.
the upstream bug has a very minimal test case using the tk shell