Re: Lilipond/nikud bug: Pango-bug ? [paps OK]

[Yotam Medini יותם מדיני]

Dov, thanks for your input.

Lilypond is a complex system. It uses Guile and Lisp-like mechanisms.
I have neither intention nor time to master these. 
In the current implementation, I cannot tell how difficult it would be
to modify Lilypond so it would interface Pango in higher level.
The main Lilypond developers may consider this approach.

Currently, I just want to try fixing the Hebrew nikud positioning with
minimal (C/C++) changes. To do this, with no Pango background, I would like to
be able to 'separate variables'. A simple tool that
+ Produces PostScript from UTF-8 script and 
  aligns Hebrew+nikud well (like paps).
+ Uses pango_itemize(..) & pango_shape() (like LilyPond)
could be very helpful.

Such tool could also be used as a test and reference program for Pango.
Is there such?

regards -- yotam

On Sat, 9 Feb 2008 20:39:31 +0200
Dov Grobgeld wrote:

> Shalom Yotam,
> 
> As you have seen the use of pango by itself does not guarantee that you will
> get the right rendering. The problem is caused by the fact that you can
> enter the pango rendering chain at different levels. paps uses the high
> level PangoLayout level. If lilypond was using that level it wouldn't have
> any problem. Nor would it need to any special processing for different
> languages and use their different backends. Especially for DejaVu sans you
> shouldn't need to do any special processing, as it in contrast to the Culmus
> fonts, contain proper opentype positioning tables for the placement of the
> nikud.
> 
> Hope this helps.
> 
> Regards,
> Dov
> 
> On 09/02/2008, Yotam Medini wrote:
> >
> > Hello Dov,
> >
> > In Lilypond (Music typesetting program) there is a bug positioning
> > diacritical marks (ni[kq]ud) with Hebrew characters. I have reported
> > the details of this bug with example in :
> >
> > http://code.google.com/p/lilypond/issues/detail?id=541
> >
> > Using the same Hebrew text using 'paps' with the same font 'DejaVu Sans'
> > gives good results. Therefore, I thought that Pango is innocent.
> > But I dived with gdb in both lilypond and paps, and I realized that
> > lilypond
> > is using Pango in a different way than paps.
> >
> > Currently, I am almost clue-less (I may learn soon ... :) about Pango.
> > But lilypond uses positioning with
> >    hebrew-fc.h: Hebrew shaper for FreeType-based backends
> >    hebrew_engine_shape( ... )
> > and it seems that paps does it in another way, although it does load
> >    modules/pango-hebrew-fc.so
> >
> > Could you give me a hint where suspects could be,
> > or different good tool to use to compare that may use
> > hebrew_engine_shape(...)
> >
> > thanks -- yotam
> >