Re: status icon AP

[Bill Haneman <bill haneman sun com>]

> On Tue, 2003-04-15 at 04:06, Bill Haneman wrote:
> > Please...
> > ! GtkStatusIcon        *gtk_status_icon_new                 (gchar *name);
> > ! GtkStatusIcon        *gtk_status_icon_new_from_pixbuf     (gchar *name,
> 
> 	You just don't get it. You absolutely cannot do this. 

Mark:

The main feature of the API above is that it provides names for icons;
the above API makes it a lot easier for developers to meet this
accessibility requirement than the alternate ATK api.  That much I *do*
get.

> If a programmer
> supplies a pixbuf, you cannot decide to override the supplied pixbuf.
> You do not know that that is the desired behaviour.

We're not talking about hijacking applications, we're talking about
providing a mechanism for theming icons.  If there's no way of making
the pixbuf icon accessible, then application developers shouldn't use
it.
 
> 	Furthermore, in future when icon themeing is in gtk+ you now have a
> completely defunct argument. How do you document @pixbuf when that
> happens? "Set this to NULL, it does nothing. It was temporarily used in
> gtk+ 2.4". What a mess.

I do not see this.  
If you are saying that gtk icon theming will supercede the status icon
API issues then we shouldn't be providing the API at all AFAICT.  But if
the .._from_pixbuf API makes sense, it will live on and you'll end up
with nameless and unthemable icons; not a good thing. I seriously
question whether we should add any API whose use is likely to result in
accessibility-impaired components.

In any case I don't see why the NAME argument would ever be useless, I
think it would always be a good thing.
 
> 	The reality is that a) for gtk+ 2.4 Owen expects icon themeing to be in
> gtk+, in which case we would have a named icon version of the API and b)
> if people use the libegg version of this for GNOME 2.4 it will probably
> have icon themeing using libgnomeui.

So why is there a _from_pixbuf or _from_file constructor at all?

-Bill
 
> Good Luck,
> Mark.
>