https://bugzilla.gnome.org/show_bug.cgi?id=784319
--- Comment #21 from Takuro Ashie <[email protected]> ---
(In reply to Martin Stransky from comment #20)
> I see that the CSD/Wayland code path is under gtk_widget_get_has_window
> (widget) condition which looks like Gtk is aware about window owned by
> container:
Sorry, you are right.
My explanation about gtk_widget_get_has_window() is incorrect.
gtk_widget_get_has_window() returns TRUE by default:
https://git.gnome.org/browse/gtk+/tree/gtk/gtkwidget.c#n8202
(priv->no_window == 0 by default)
In addition, when gtk_widget_get_has_window() is FALSE, a parent GdkWindow is
used for the widget's GdkWindow, it doesn't create its own GdkWindow.
On the other hand the explanation about "unrealize" is correct.
I confirmed that it's not called so that no one destroy the GdkWindow of
MozContainer. When MOZ_WAYLAND isn't defined, it's destroyed by the following
line:
https://git.gnome.org/browse/gtk+/tree/gtk/gtkwidget.c#n10589
But when I address it, sometimes another similar crash occurs.
It's more rare than before...
We should continue it at https://bugzilla.redhat.com/show_bug.cgi?id=1467104.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
wayland-bugs mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/wayland-bugs
