Re: GType::register RFE

["muppet" <scott asofyet org>]

i love replying to myself, it makes me look crazy.

muppet said:
> i don't dispute your need to be able to run stuff at class_init time --- what
> i'm saying is that there's no need to put that stuff into a function, because
> any code placed immediately after Glib::Type::register will run after the
> class has been initialized.

to clarify, according to
http://developer.gnome.org/doc/API/2.0/gobject/gobject-GType.html#GClassInitFunc
, the class_init runs last, that is, after your base_inits --- so as long as
you execute your class_init stuff before you go on to create more classes that
may inherit from it, you're fine.  with this, the typical setup of a perl
module is fine.

sorry to beat a dead horse --- that's probably the fifth time i've said it,
but i feel like it's not coming out the way i want to.


>  (and if not, you can force it to exist when you
> get into an xsub called there.)

in fact, to prove this to myself, i wrote an extra little xsub in Glib::Type:

  void
  do_something_strange_and_unnatural (class, const char * package)
     PREINIT:
        GType gtype;
        GTypeClass * class;
     CODE:
        gtype = gperl_type_from_package (package);
        class = g_type_class_peek (gtype);
        warn ("  %s => %s %d\n", package, g_type_name (gtype), gtype);
        warn ("  0x%p => %s\n", class, g_type_name (G_TYPE_FROM_CLASS (class)));

then i called this from various INIT_BASE and post-registration spots, and
never got anything i didn't expect.


> an implementation of that algorithm, and a sample program, and a t-directory
> test version of the sample are attached.

looks like it contained a vestigal INIT_CLASS that won't be called.  bad
muppet, no cookie.


-- 
muppet <scott at asofyet dot org>