Re: AtkState

["Padraig O'Briain" <Padraig Obriain Sun COM>]

> X-Sender: timj birgrave birnet private
> To: "Padraig O'Briain" <Padraig Obriain sun com>
> cc: otaylor redhat com, Havoc Pennington <hp redhat com>, 
gnome-accessibility-list gnome org
> Subject: Re: AtkState
> MIME-Version: 1.0
> 
> On Tue, 1 May 2001, Padraig O'Briain wrote:
> 
> > > 
> > >  * I think the way AtkState currently works with the extensible
> > >    enum and:
> > > 
> > >    typedef guint64 AtkStateMask
> > >    #define ATK_STATE(state_enum)   ((AtkStateMask)(1 << 
> > ((guint64)(state_enum)%64)))
> > > 
> > >    Is going to cause fits for language bindings. It probably
> > >    would be better to simply have functions to get/set a
> > >    list/array of AtkState for the AtkObject.
> > > 
> > 
> > I propose to change 
> > 
> > AtkState                atk_object_get_state                      (AtkObject 
*ac
> > cessible);
> > 
> > to
> > 
> > AtkStateSet             atk_object_get_state_set                  (AtkObject 
*ac
> > cessible);
> > 
> > and define the AtkStateSet interface as in the attached file.
> > 
> > Hopefully, this interface will work for language bindings.
> 
> 
> ok, i can't say i understand your "states" enough to judge whether exporting
> the flags into a flyweight interface is a good idea. but in any case, since
> with the new approach you return a new object, you prolly meant to write:
> 
> AtkStateSet* atk_object_ref_state_set (AtkObject *accesible);
> 
> right?
> 

Right, thanks.

Padraig

> > 
> > Padraig
> > 
> 
> ---
> ciaoTJ
>