Re: PangoXft and PangoFT2 patch from hell

[Owen Taylor <otaylor redhat com>]

Hans Breuer <hans breuer org> writes:

> >> > Anybody have a simple (but not
> >> > too simple) working test program for pangoft2?
> >> > [...]
> > [...]
> >Perhaps there could be a viewer-ft2-win32.c that uses GDI
> >calls directly for drawing... 
> This was the keyword :-) 
> Not sure if it isn't too simple but I've written testfonts.c
> for exactly the purpose to debug pango without gtk. It is in
> cvs and at least works with pangowin32. Porting it to use
> FT2 should be a matter only a few #ifdefs ...
> 
> > > Perhaps we should restore viewer-ft2.c.
> > 
> > viewer-ft2.c is fundementally unfixable because it depends on 
> > having copy of GTK+ that you can link to the same GLib as
> > Pango but that doesn't require Pango.
> Sorry for my ignorance but I fail to see the problem.
> It obviously won't work with a totally broken Pango but
> could be of some help with Pango having smaller glitches.

What I'm saying is that viewer-ft.c *using GTK+* doesn't
make any sense because:

 GTK+-1.2 requires GLib-1.2 which conflicts with GLib-2.0,
   which is required by Pango.
 GTK+-1.3.x requires Pango so can't be used in a Pango test case

When Tor wrote this, I believe he was relying on the 
fact that win32-production GTK+ could be linked against
HEAD GLib, but that has never worked on Unix, and probably
no longer works Win32.

Regards,
                                        Owen