[XeTeX] XeTeX and Classical Mongolian

[Jonathan Kew]

On 3 Oct 2004, at 11:57 pm, Thomas T. Pedersen wrote:

> Has anyone tried to use XeTeX with Classical Mongolian?

You're probably the first!

> Code2000 vers. 1.14 has OpenType support for Classical Mongolian and 
> works with BabelPad under Windows XP with Uniscribe 1.471.4063.0.
>
> In XeTeX, i
> f I just define the font as "Code2000:script=mong", the output 
> displays the isolated form of the characters.
>
> I've experienced with three available features 'init', 'medi' and 
> 'fina', adding e.g. ";+init;+medi;+fina". The output changes but it's 
> apparently using either the initial, medial or final forms of the 
> characters - and not the right combination of them.
>
> Attached you'll find a short sample text string in Classical Mongolian 
> to experiment with.
>
> Any suggestions to make it work are welcome.

I'm afraid it just won't work, for now. There's no Mongolian script 
shaping engine in XeTeX at this point (so "script=mong" is ignored). 
And the default engine used for Latin, etc., won't work because it 
applies the features you specify to *all* the glyphs. But for Mongolian 
you need the engine to understand the script and apply *different* 
features to each glyph depending on its context.

XeTeX is using the ICU layout library 
<http://oss.software.ibm.com/icu/> for OpenType support, so you could 
file a bug report/feature request regarding Mongolian support there, 
and it may eventually become available.

Sorry to disappoint you. If there's an AAT-enabled Mongolian font 
available, that would work, as the AAT model puts all the behavior into 
the font tables, requiring no script-specific support in the 
application.

(Any volunteers to write the OT shaping code? ICU is open-source.....)

Jonathan