Re: accelerators

[Tim Janik <timj gtk org>]

On 1 Nov 2001, Havoc Pennington wrote:

> 
> James Henstridge <james daa com au> writes:
> > Is it enough to be able to attach accelerators to signals of objects
> > not derived from GtkWidget?  That is already possible with the current
> > API (I made use of it in that menu actions demo I wrote a few months
> > back), so preserving this ability might be sufficient.
> 
> I don't think so, the main reason for allowing callbacks is so that
> app authors can easily add an accelerator, without having to 
> do a key_press_event handler. Any API that requires deriving a GObject
> is essentially nonexistent for many app authors.

erm, a couple comments for the sake of clarity:
1) key handlers that apply to all widgets of a certain type, e.g. entries
   should be done via binding sets
2) adding a key handler to an object does _not_ require derivation,
   connecting to ::key_press_event is sufficient.
3) accelerators are gtkwindow global, so can't be viewed as simply
   key_press_event handlers on widgets (or you mean deriving GtkWindow
   in the above, but then that'd mean extra derivation for each enitity
   that adds a shortcut ;-[] )
4) setting up closures as accelerators, takes away two significant burdens
   for a shortcut:
   - the programmer doesn't need a specific obejct that handles this accel
   - the programmer doesn't need the accel be routed through a specific
     object's signal (i.e. add multiple RUN_ACTION signals to an object
     just to handle  couple different accelerators)

> 
> Havoc
> 

---
ciaoTJ