Re: [gtkmm] Event signals should connect _before_?

[Chris Vine <chris cvine freeserve co uk>]

On Friday 14 November 2003 3:02 pm, Murray Cumming Comneon com wrote:
> What do people think of this idea?
> http://bugzilla.gnome.org/show_bug.cgi?id=126213
> (Please see the dependent bugs for actual examples of the problem.)
>
> I am beginning to suspect that every single X-event signal handler
> ( http://www.gtkmm.org/gtkmm2/docs/tutorial/html/apbs06.html )
> should connect before (or connect_notify) in order to actually work.
>
> I would like to hear about examples where this would not "just work". If I
> don't hear of any examples then I might release a gtkmm 2.3.x version with
> it and see if it breaks your applications. Not that many of you are using
> gtkmm 2.3 with your applications now.

As you say, it is difficult to see how a default connect after could work with 
X event handlers which return true, if the return value of true is intended 
to stop further handling of the event.  I have been checking some of my code 
and the only cases where it is relevant use the equivalent virtual protected 
methods, which I always write so that they do their own ordering, so I cannot 
help you beyond that.

I haven't looked at the signal proxy code for X events, but I assume that with 
a connect before (rather than a connect_notify), a return value of true would 
actually stop any further handling of the event?  Otherwise changing the 
default would not achieve very much.

It is quite surprising that this hasn't cropped up before.

Chris.