Re: GTK--: hello.cc questions

[Owen Taylor <otaylor redhat com>]

Tero Pulkkinen <terop@students.cc.tut.fi> writes:

> Owen Taylor <otaylor@redhat.com> writes:
> > There is no "destroy_event". 
> 
> I find this surpricing...
> 
> in gtk-- I have the line:
>    SIGNAL_SPEC(gint destroy_event(GdkEventAny*));
> 
> This line to NOT give a compile time error, it requires that there exists
> a destroy_event virtual function.
> 
> I thought all the virtual functions in gtk+ has corresponding signal available
> too :) (thats what gtk-- expects.. but of course that cannot be checked in
> compile time as the signals in C are identified with string identified :)
>
> Also there is no equivalent of plain destroy in gtk-- and it still seems to
> compile... really odd :)

Well, not that odd. As you've no doubt figured out, I was
suffering a momentary memory lapse.

There is a GDK_DESTROY event, and a corresponding signal.
This GDK event is received when the server destroys a window.

But, if the application receives this, it means trouble,
since it means that some other application running on
the X server has gone ahead and destroyed one of your
windows. There is no way to receive one of these critters
in a GTK+ window, since you are always supposed to set
user_data to NULL on a window before destroying it,
and in that case, no events will be forwarded.

So, yes, there is one, but it is completely useless from
the GTK+ and (gtk--) point of view.

Regards,
                                        Owen