Re: [Vala] Implicit lamdas/closures

["A. Walton" <awalton gnome org>]

2008/9/16 Christian Hergert <christian hergert gmail com>:
> This sounds particularly interesting to implement.  As we would almost
> certainly need some sort of "Async Handle" to work with.  .NET has
> IAsyncResult, Twisted has Deferred, etc.  What makes this tough is that it
> would require some sort of addition to GLib unless we were to require an
> extra library like Gee.  However, I guess we could generate a special struct
> that is used for this handle.

GLib has GAsyncResult which likewise follows the IAsyncResult pattern,
in the GIO library. Unfortunately as there isn't a bottom library for
such items (GObjects and interfaces with general usefulness), it had
to be stuck in GIO. But it also makes sense as GIO itself desperately
needed a way to generalize asynchronous programming.

-A. Walton

> Of course, I could be totally wrong here and someone may have a clever way
> around this.  I'll let it stew in my head over night; anyone else have
> thoughts on this?
>
> -- Christian
>
> On Tue, Sep 16, 2008 at 4:28 PM, Samuel Cormier-Iijima <sciyoshi gmail com>
> wrote:
> > On Tue, Sep 16, 2008 at 6:43 PM, Jürg Billeter <j bitron ch> wrote:
> > > On Tue, 2008-09-16 at 15:39 -0700, Christian Hergert wrote:
> > > > I absolutely love the yield idea, very python-twisted'ish.  Lets make
> > > > sure the yield can return a value as well so we can have the
> > > > following:
> > > >
> > > > var value = yield obj.do_something ();
> > >
> > > Sure, return values should certainly be possible.
> >
> > I'd love to see this in Vala - it would make asynchronous programming
> > much nicer. Would there be a syntax for specifying what types can be
> > sent in to the iterator/coroutine (assuming that the yield syntax
> > could be used for general iterators), or would yield always have to be
> > used with the _async/_finish API (and value's type inferred from the
> > sync version of the call)? Also, what would the C API look like for
> > these - functions that return GClosures or automatically-generated
> > sync/async versions of the function? I'm guessing that Vala could
> > automatically determine which one to do if it sees that you're
> > returning the result of a method marked as async-able, or maybe there
> > should be different keywords for either case... For example:
> >
> > string do_something(Socket sock) {
> >    try {
> >        yield sock.connect();
> >    } catch (Error e) {
> >        return null;
> >    }
> >
> >    var data = yield socket.read(5);
> >
> >    return data;
> > }
> >
> > Might be nice if this generated do_something, do_something_async, and
> > do_something_finish automatically.
> >
> > --
> > Samuel Cormier-Iijima <sciyoshi gmail com>
>
>
> _______________________________________________
> Vala-list mailing list
> Vala-list gnome org
> http://mail.gnome.org/mailman/listinfo/vala-list