Re: GSignal documentation

[Tim Janik <timj gtk org>]

On Mon, 20 Aug 2001, Joshua N Pritikin wrote:

> On Mon, Aug 20, 2001 at 02:08:39PM +0200, Tim Janik wrote:
> > On Mon, 20 Aug 2001, Joshua N Pritikin wrote:
> > > Perhaps what i am trying to say is that the signal's itype/instance only
> > > determine the signal's scope.  The emitters and receivers of a signal
> > > may or may not be part of the signal's instance.  To contrast, the
> > > documentation seems to imply that the signal is probably emitted by
> > > the instance.
> > 
> > which is correct.
> > a signal usually is created in the class_init function of a certain
> > instance type, and later it is emitted for instances of (or derived
> > from) that type.
> 
> i'm not restricting myself that way.  i can just as easily create signals
> in an arbitrary scope class, instantiate it, and logically connect/emit

i'm not sure what you mean with "scope class", do you mean an instance class or
not?

> signals from unrelated emitters and listeners.  In fact, i believe the
> signal system should be *described* as this less restrictive model.  Again,
> this is just a matter of documentation.  Otherwise the API & etc is fine.

if you're emitting signals not from within the instance implementation,
you should flag it as G_SIGNAL_ACTION, we use such signals as "freely
invokable methods" on an object, e.g. for key bindings in gtk.

---
ciaoTJ