Re: Marshalling CClosures

[Havoc Pennington <hp redhat com>]

Tim Janik <timj gtk org> writes: 
> havoc gave the wrong impression by indicating that closures are only
> usable from signals, we already support what you want.
> 

I didn't say closures are only usable from signals - I said we should
have some prebuilt feature for when you want to have a reentrant list
of handlers supporting connect/disconnect/emit, and that feature would
automatically handle the marshalling details. (You would pass in the
marshaller on creation of the handler list object.)

I believe your argument has been that signals are exactly this
feature, i.e. if you want the reentrant handler list you should use
the signal system.

As you have pointed out, the marshaller issue is only 1/10 the battle;
there's still a bunch of stuff Gergo has to implement to get this
right, re: reentrancy etc.

In any case, too late to change for 2.0.

Havoc