Re: [gtkmm] on_realize()!=signal_realize() ? , was: glade-2 + on_realize()

[Christof Petig <christof petig-baender de>]

Silviu D Minut schrieb:
> I assume I'm doing something wrong (e.g. use the wrong return value), but
> that's because of the code generated by glade. Please take a look at the
> code below, I really need help with this.

return value? signal_realize needs a void (void) callback. That's no 
return value at all ...

> > It's wrapped like every other signal. Because this is an X event signal
> > http://www.gtkmm.org/gtkmm2/docs/tutorial/html/apbs06.html
> > he's probably just using the wrong return value, or needs to use
> > connect_notify() instead.

Quoting widget.h:

Glib::SignalProxy0<void> signal_realize();

Is this an X event signal - if so, why does it have neither argument nor 
return value?


> If I understand this, the problem is with
>
> void on_drawingarea1_realize();    // note the void return type
>
> and/or
>
>    drawingarea1->signal_realize().connect(SigC::slot(*this,
> &window1_glade::on_drawingarea1_realize), true);
>
> These were created by glade. I'm not supposed to edit the files where
> these two instructions occur, because glade overrides them. Is glade
> generating the wrong code? Is there any way to make it do the right
> thing?

I don't think glade is generating wrong code

> Also, I noticed that if I make on_drawingarea1_realize() a nonvirtual
> function in window1_glade and implement it, it works.

Oooops. Well this clears it out:

virtual functions are only legally called after the class is fully 
initialized. So if for any reason ;-) on_realize is called within 
window1::window1 it is not allowed to call 
window1::on_drawingarea1_realize before the ctor has finished
[window1_glade::on_drawingarea1_realize is legal].

Perhaps putting drawingarea1 in a "separate class" will help here 
(because then drawingarea1::drawingarea1 has finished before the widget 
will be realized).

All in all this is just a bit beyond the border of glade(mm)'s 
capabilities/design. I never realized ;-) this.

   Christof