Re: [Vala] interface properties: unexistent accessor

["Alexander Bokovoy" <ab altlinux org>]

On Mon, Nov 17, 2008 at 11:35 AM, Ildar Mulyukov
<ildar users sourceforge net> wrote:
> > > > I discovered the problem (and created this simple test case). The root
> > > > of it is that the unexistent accessor referenced from the implementor class.
> > > > Maybe it's better not to do it?
> > >
> > > Which vala build is it? (I've just pushed r2023 with a lot of fixes to
> > > Sisyphus).
> >
> > Discussed with Ildar. Apparently, main problem is that he has interface
> > definition and a class that implements interface in separate vala source
> > files. During generation phase valac puts ns1_interface1_set_name() into one
> > .c file and marks it as static. Therefore it is not accessible from the
> > second .c file where ns1_class1_set_property() is defined.
> >
> > In such cases _set_property() functions generated by
> > gobject/valaobjectmodule.vala:add_set_property_function() need to have
> > indirected calls via ns1_class_real_set_name(), not
> > ns1_interface1_set_name() function.
> >
> > Experimental patch attached.
>
> Your speed is awesome!
> But are you sure it would work when an interface would be in a library and
> the class will link to this library? I'm not sure about this.
As this patch doesn't rely on internals of interface it works fine
with externally provided interfaces. I tested that with AsyncResult
from gio-2.0.vapi.
-- 
/ Alexander Bokovoy