Re: pango, first try

[Werner LEMBERG]

> I had a look at using Pango for typesetting text. See the attached
> program.

Which version of pango do I need for that?  With 1.2.5 I get a
segfault with G_OBJECT_TYPE_NAME...  Anyway, after commenting out the
printf statement I was able to proceed.

> I am wondering: are the number codes (printed as c XXX) really
> unicode codes? When I convert
> 
>       \'all
> 
> as utf-8, I get 
> 
>       c 67 w 14336 x 0 y 0
>       c 177 w 6144 x 0 y 0
>       c 177 w 6144 x 0 y 0
>       c 0 w 0 x 0 y 0

These are glyph indices into the font which Pango has selected.  I
used `strace' to find out which font Pango actually was using.

> In any event, I will need a table of unicode -> glyphname mappings.

Use FreeType's FT_Get_Glyph_Name to get a glyph name for a given glyph
index.  Pango has probably a similar function.  For the moment, we
should restrict ourselves to PS flavoured fonts which can be easily
integrated into a PS output.

> * PangoGlyphUnit <-> Real World dimension conversion.

Where's the problem?  Everything should be converted to PS points,
right?

> * How to integrate fontconfig into lilypond font-selection?

This is an important question, but it should be delayed.  For the
moment I suggest to directly use the PS font name.

> * How to handle dichotomy with TeX fonts and OTFs?

What exactly do you mean?

> * Do we need any settings for language/shaper?

Yes.  LilyPond should give access to OpenType features by selecting
the script, the language, and the features.  Think of Turkish where
you must not use the `fi' ligature.


    Werner