Re: Button patch (Was: enhancing GTK visual quality)

[Damon Chaplin <damon helixcode com>]

Tim Janik wrote:

> hm, except for you using a timeout, the patch looks pretty sane.
> what i don't like about using a timeout is that if e.g. a ::clicked
> handler calls gtk_main() recursively, the button becomes depressed
> after 100ms when activate via the keyboard, while it wouldn't do
> that for mouse events.
> i think we should be consistent there with cleaning up the ACTIVATED
> state at the end of the signal emission (for 1.2.8, we'd have to do
> that in an _after() handler which may impose problems for stopped
> emission, but the patch has to be written against CVS HEAD anyways,
> and there we'll have something like a cleanup class handler).

I think you have to use a timeout really.
If you restore the button's state to NORMAL/PRELIGHT after the "clicked"
emission finishes, you may not even see the button pressed in - it
may all happen too fast.

(I did miss one thing in the patch - if the button has been set to
insensitive it should leave it in that state.)

Damon