Re: Pluggable widget types and implementations

[Bill Haneman <Bill Haneman Sun COM>]

Tim Janik wrote:
> On Sat, 9 Dec 2006, Bill Haneman wrote:
>
> > Hi All;
> >
> > As I understand it, the proposal below would probably break gail
> > unless/until we roll it into gtk+.
>
> can you please elaborate why this should be the case?
> basically, the proprosal is about exchanging widget types. 
My understanding is that the pluggable type system would allow for 
alternate implementations for a given GType.

If that is the case, then the implementation code for, say, a 
GtkTreeView might differ in a 'pluggable' environment from the stock 
gtk+ implementation treeview code. If this is the case, the gail code 
will almost surely break; this is because, while gail uses public gtk+ 
API for its support (thus the API would be the same in a pluggable 
environment), gail does currently depend on implementation/behavior 
details of stock gtk+ widgets (i.e. order and type of children, perhaps 
signal emission order, etc.).
> as long
> as widgets are supported by gail, nothing should break. and if they
> aren't, we're simply talking about accessibility TODOs and the
> pluggability doesn't introduce any *new* breakage.
>
> accesibility and pluggability are simply orthogonal.
Not sure I agree. But then, perhaps I have the wrong concept of what is 
actually intended by a 'pluggable' type system. If so, please feel free 
to enlighten me (on or off-list).

Bill
> > Bill
>
> ---
> ciaoTJ