Re: Bypassing Force Quit

["Brian J. Tarricone" <bjt23 cornell edu>]

On Wed, 16 May 2007 20:54:34 +0200 David NeÄas (Yeti) wrote:

> On Wed, May 16, 2007 at 11:28:55AM -0700, Brian J. Tarricone wrote:
>
> > This is caused by the window manager, not gtk.  It thinks the
> > application isn't responding because it takes too long to return from
> > the delete-event handler.
>
> This is interesting, do you know what exactly the window
> manager is waiting for?  Because it has of course no idea
> what your code does inside (namely whether it is in some
> handler or not) -- and you let Gtk+ main loop run, so it
> responds to events normally.  Therefore the window manager
> expects some specific reaction from the client on delete.
>
> However, it cannot be destruction of the window as that
> would break all applications that allow one to cancel the
> closing of the window.  Returning TRUE also prevents the
> default Gtk+ handler from running, so anything else Gtk+
> could do there cannot be the thing.  So what is it?

My assumption here is that the WM was sending a _NET_WM_PING to the
window after trying to close it (as suggested in the WM spec[1]), but
you're right, presumably gdk/gtk should still be responding to those
events even with a modal dialog up.  Unless the recursive mainloop
created by gtk_dialog_run() for a modal dialog *wouldn't* allow gdk/gtk
to respond to a ping on a different window.  Not sure, but that's the
only reason I can see: otherwise I don't know why this would be
happening, as I'm pretty sure that you're right that returning TRUE or
FALSE from the gtk event handler doesn't actually send any X events.

        -brian

[1]
http://standards.freedesktop.org/wm-spec/wm-spec-latest.html#id2512097