Re: more mainloops in more threads?

[Paul Davis <paul linuxaudiosystems com>]

On Tue, 2008-04-01 at 23:17 +0200, Roberto -MadBob- Guido wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Tuesday 01 April 2008 22:39:15 Paul Davis wrote:
> > it might help to realize that signals are nothing more (or less) than a
> > way to execute a list of closures
> > 
> Exactly as I was afraid to be.
> 
> So, adding a signal connection adds an item into that (unique for all the 
> application) list, increasing time to lookup and traversing. Right?

Wrong. The list of closures is attached to that *signal*, which in turn
is a property of the object that emits it.

> And, as I can understand, there is no way to manage different lists of 
> handlers for different contexts (say: for different threads).

as chris pointed out, your new understanding is wrong ...

> I think Glib will need some low level revisions to match requirements for near 
> future computation models...

i think i would advise reading a bit more and thinking about some of the
projects that already exist and their likely computational needs before
launching into such a far-reaching condemnation.

multithreaded programming is not a new thing. many glib-using
applications are heavily (and occasionally massively) multithreaded.