Re: [gtk-list] Gtk-- connect() woes (was Re: gtk-- information ??)

[Tero Pulkkinen <terop students cc tut fi>]

Jon Trowbridge <trow@emccta.com> writes:
ÿ> I was not able, however, to get signals whose callbacks take arguments to
> work properly.  I took the following code snippet:

[SNIP]

>   connect(ebox.button_press_event, button_press_callback_fn);

Yeah, I think you need to have
   connect(ebox.button_press_event, &button_press_callback_fn);

hmm, I think I might have many examples in the web page without the extra
&.:) I'll check them all ... :)

> no matching function for call to 
> `connect (Signal_proxy1<int,Gtk_Widget,_GdkEventButton *> &,
>           void ()(_GdkEventButton *))'
> 
> /usr/local/include/gtk--sig.h:735:
> candidates are:
> connect(Signal0 &, void (*)())
> 
> /usr/local/include/gtk--sig.h:1172:
> connect<int, Gtk_Widget, void ()(_GdkEventButton *), _GdkEventButton *>
> (Signal_proxy1<int,Gtk_Widget,_GdkEventButton *> &,
> void ( &)(_GdkEventButton *))
> 
> I tried looking at gtk--sig.h line 1172, but it seems to involve
> macro-expansion magic and I ran screaming in the other direction...
> 
> But notice what it is suggesting as a match: a connect() whose second arg
> is a void ( &)(_GdkEventButton *)).  If I'm not mistaken, isn't that
> a *reference* to a void function taking a _GdkEventButton* as an argument?

Yes. This is called "function object" in the STL for example. And templates
do make a difference between reference/pointer.

Hmm, I wonder why that doesnt match the function object -version of connect()..
It should. (tho those are supposed to be used with &...)

> And there is certainly no mention that I can find of references to functions
> in the latest edition of Stroustrup.

It can only happen in template expansion. Templates are special - they do
always want perfect match of the signature.

-- 
-- Tero Pulkkinen -- terop@modeemi.cs.tut.fi --