Re: pango-0.9 arabic module

[Owen Taylor <otaylor redhat com>]

Karl Koehler <koehler@or.uni-bonn.de> writes:

> Now I checked out the CVS version on saturday and found there still 
> is some trouble:
> 
>   -  I tend to get Messages as the following:
>      
>      ** CRITICAL **: file pango-layout.c: line 839 (pango_layout_get_cursor_pos): assertion `index >= 0 && index <= layout->length' failed.
>      
>      ( a *lot* of them !! )

This should be fixed now, I hope. I made a couple of big check-ins
last night to Pango and GTK+ last night with improvements to
editing behavior and fixes.

In particular, the new text widget is working pretty well now, except
for performance, and a few remaining bugs. (There is a separate
testtext program to try it out.) Cursor motion, selection, etc, in
that should all work just about correctly.

(Actually, if you want to check out the block of Arabic text in
testtext, and make sure that it isn't saying anything inappropriate,
I'd appreciate it. I took it out of a article in UTF-8 I found
somwhere, but I lost the original reference, so I forget what it was
about. Which makes me a bit nervous. I think it was a technical
article of some sort.)

>  -   The following  Message still occurs often:
>      ** WARNING **: Invalid subfont 0

Hmmm, I'm not quite sure why you are getting this. Basically, 
the only time I can think of that this would get produced
is if pango_x_get_unknown_glyph() completely fails, which
shouldn't happen under any normal circumstance.

Maybe you can put a break on the last line of this function,
(the 'return 0') and see investigate (what is xfont->fonts[0]).
 
>  -   Sometimes gtk-test segfaults and I can't see the reason.
         
That may be related to the above problem. But if it continues,
and you can get a backtrace, send it along.
 
>  -   Make-install of the new gtk+ failes to install properly,
>      causing configure-scripts testing for gtk+ to fail.
> 
> But it is awesome to see all ( most of ) the labels rendered correctly,
> and to be able to enter at least some arabic text !
> Things seem to work much better than which the (old) snapshot-downloads.

Yes, I've made a lot of fixes. I'm going to try to do another set
of snapshots later today.

> I now (finally ) added support for the langbox--6.8x font-encoding;
> with the proper aliases there should be less problems with fonts now.
> You can download my new version of the arabic shaper from
>   http://titan.cs.uni-bonn.de/~koehlerk/u/

OK, great. I'll put it in. 

(If you ever want direct CVS access to make changes, let me know. I
seem to remember that you had firewall problems, but if that changes /
has changed, I'd be happy to give you an account.)

                                        Owen