Re: Pango Status Report, 4 Apr 2000

[Owen Taylor <otaylor redhat com>]

Pablo Saratxaga <pablo@mandrakesoft.com> writes:

> Kaixo!
> 
> On Wed, Apr 05, 2000 at 12:40:42AM -0400, Owen Taylor wrote:
>  
> >  My project of switching GTK+ over to use Pango is finally
> >  beginning to show some results. Some first screenshots are
> >  at:
> > 
> >   http://www.pango.org/gallery.shtml
> 
> Really awesome; and in particular the fact that widgets are mirrorable
> is a very nice thing; I wasn't aware that it had been thought when
> building gtk+, big kudos for it ! 

I don't know if it was thought about when building GTK+, but the design
of layout in GTK+ turns out to work very well for this.

> (very few is missign to have perfect
> behaviour, I spotted the dropdown lists, where the 'handle' is not mirrored;
> the tabs, where the tab should be put at the right of each sheet;
> and I wonder if Window Managers can be made aware of that (maybe a new
> WM hint would be needed to tell that a window is in RTL mode (buttons
> are mirroredand the title right justified); in other words what is still
> missing are widgets that need to be redrawn differently and not only put
> at another palce, when changed to RTL mode)

Yes, there are some more widgets that need to be done. The current
widgets were all done in about 2-3 hours, so it shouldn't be all
that much work to finish off the rest.

Window manager support is something I haven't even thought about yet. 

> >  Karl Koehler made a large number of improvements to his Arabic
> >  shaper, including expanding the available ligatures, supporting more
> >  font encodings, and adding Farsi support. (Which apparently
> >  needs testing.)
> 
> I would like to help with font support. Is there some small tutorial
> available ? Otherwise, should I start looking at pango code or libunicode
> code ?

There is no real documentation on writing a shaper yet. Which, I
think, would be the documentation you would need. So, it could be said
that its pretty amazing that we actually have 4 contributed shapers
now (Tamil, Korean, Arabic, and just added in CVS, Devanagari.)

I need to start writing this documentation soon.
 
> >    - Come up with a XKB-based keyboard map for Arabic, so people
> 
> Is XKB mandatory ? I run Xserver with XKB support disabled and I load
> an xmodmap file, but when I tryed typing (the edit widget of the tests
> of gtk+ 1.3) it outputs errors of "illegal utf8 string" (or something like
> that, I don't remember exactly); I use keysysms in my xmodmap file
> (eg: agrave, ccdeilla,...) and I'm with LC_ALL set to a latin1 locale.
> Or is it that those keysysms have yet to be defined ? (where?)

XKB shouldn't be necessary now - in fact, when I was testing out entering
Hebrew characters, I was doing that with xmodmap. But the version in
the snapshots doesn't handle anything but hebrew and ascii very well.

The version of GTK+ in CVS should work a lot better. (I'll make another
snapshot soon.)

There _are_ things that we can do with Xkb that we can't do with
xmodmap - in particular, we can monitor the current group of the
keyboard. So its possible that there will be optional xkb support
in the future. But I would expect things will continue to work
at least minimally with xmodmap in the future.

> On another side, I know several people that are willing to do translations
> to Arabic/Hebrew/Farsi, but that untill now I told them it wasn't possible
> due to lack of support. Do you think it would be a good idea to create
> some complied versions of gtk+-with-pango (of course with BIG LETTERS warnings
> telling that gtk+ 1.3 is development only etc); then they can start doing
> translations; that will have several benefits: when gtk+ 1.4 will be out
> it will already have translations for those languages from first day;
> pango/gtk+ developpers will have real texts to test with; and also a real
> world testbed with feedback from native people that can tell interesting
> things about some details that non native people can maybe overseen.   

Yes, I think this is a reasonable idea. In fact, it was partly so that
native speakers could start trying things out that I posted the
tarball snapshots of GTK+ with compilation instructions last week.

In a few days, I hope to have a new release with a working Pango-ized
multiline text widget. At that point, I might add snapshot RPMS
(installing in /opt/pango) to the snapshot tarballs. 

The multiline text widget should make it easy to build a tool people can
use to edit translations.

Regards,
                                        Owen