Re: GnomeClient replacement?

[Bill Haneman <Bill Haneman Sun COM>]

I have to disagree.  As I recall the history, it was the
GTK_MODULES/gtk+ fix that caught most of the flak (and still does,
relying as it does on an env variable).  At the time that the gconf key
for assistive technology support was first introduced, nobody called it
a hack.

Now that we have xsettings in gtk+ I agree though that an XSETTING makes
more sense.

Bill

On Wed, 2006-07-19 at 18:34, Havoc Pennington wrote:
> Bill Haneman wrote:
> > gnome_program_init also loads the accessibility support, calling gconf
> > in the process.  It's not clear to me that this could conveniently be
> > put elsewhere without complicating the dependencies of other modules...
> > 
> 
> This is a broken hack that should have been killed long ago, should 
> never have been allowed into libgnome at all since it means that 
> gtk-only apps need to either cut-and-paste the code or just not be 
> accessible, despite having all the other a11y code already in gtk.
> 
> It can be done with e.g. a GTK_MODULE instead, iirc. or perhaps an 
> xsetting type of deal to get the gconf flag into gtk itself.
> 
> I remember threatening long ago to kill the cut-and-paste from metacity 
> after a release or two if nobody fixed this properly; Elijah, you should 
> do that, I think it's been maybe two years or more! Fortunately Elijah 
> is probably nicer than me and won't enforce the threat ;-)
> 
> There's a metacity bug about it iirc.
> 
> This has got to be a perfect poster child for why maintainers should 
> reject broken half-measures on grounds that nobody will fix it properly 
> once you accept a broken patch.
> 
> Havoc
>