Re: g_value_ types ...

[Michael Meeks <michael ximian com>]

Hi Tim,

On Thu, 3 May 2001, Tim Janik wrote:
> for corba that's probably usefull, though the implementation shouldn't
> go into glib. 
 
        Sure :-) I'm just asking for some reserved type ids.
 
> i can just preserve a fundamental type id for you: G_TYPE_CORBA(_ANY)
> that you can use for GValues, and the implementation for that then
> comes with orbit or bonobo. you could however implement that as a  
> BOXED type also. since i don't know too much about CORBA, i'd like to
> hear owen's opinion on that (hope you don't mind me Cc-ing him).   

        Owen, can we beg some reserved type ids for a legacy CORBA base
system ? ;-), perhaps in the spirt of greater scrutiny, etc. etc. I'll CC
gtk-devel too - though this means I'll never notice the reply.

> G_VALUE_BONOBO_OBJECT doesn't make much sense though, from what i
> know, you're deriving BONOBO_TYPE_OBJECT from G_TYPE_OBJECT already  
> right? so you can use that in GValues right away (GValues figure it's
> a G_TYPE_OBJECT and know how to handle that).

        Ok, well there are 3 things that might be useful for me:
        
        G_TYPE_CORBA_ANY - containing a pointer to an any [ of course ]
        
        G_TYPE_CORBA_OBJECT - containing a CORBA_Object reference, copy
                / free would use CORBA_Object_duplicate / release
        
        G_TYPE_BONOBO_OBJREF - also containing a CORBA_Object reference
                but in this case to a remote Bonobo_Unknown, copy / free
                would use Bonobo_Unknown_ref / unref etc.

> > Currently using a G_TYPE_POINTER for the GValues Martin has setup,
> > but possibly this is less than optimal.
> 
> G_TYPE_POINTER can be considered a nuklear waepon against language
> bindings ;)

        :-)
           
        Regards,
  
                Michael.
-- 
 mmeeks gnu org  <><, Pseudo Engineer, itinerant idiot