Re: Non-BDF Devanagari Support in Pango

[Owen Taylor <otaylor redhat com>]

Mark Leisher <mleisher crl nmsu edu> writes:

>     Chookij> Does this approach mean that Pango is still running in OTF
>     Chookij> support for hindi, not TTF support for hindi, right ?
> 
> Yes, this would mean assuming that all Devanagari OTF fonts have Hindi
> rendering included and that Pango would only work with OTF fonts.
> 
> As you are no doubt well aware, there are lots of TTF Devanagari fonts out
> there, most of which have different sets of glyphs.  Each of these will need a
> rendering program built for them before they can be effectively used.
> 
> I haven't really paid much attention to discussions about this recently, but I
> believe both Owen and Robert are very aware of the situation and Pango will
> eventually have at least some mechanism for adding TTF (and bitmap) fonts, if
> one is not already in there.

Actually, this points out a more important problem - to use existing 
TTF devanagari fonts, it isn't just a matter of getting the ligature
data, but dealing with the fact that Devanagari fonts are typically
in some custom encoding made up by rolling dice.

So, I suppose what we need is not just a way of storing ligature data
client side, but also storing a mapping table client side, that maps
the code points of the font into the code points of a sane encoding.
(I believe that for the BDF fonts, Robert used the Unicode encoding
with some extensions to deal with glyphs that don't correspond
exactly to characters.)

So, we might want a configuration file that looked something like:

[XLFD_PATTERN]
map_file: foo.map
ligatures: [ligature_hack data]

(I'm not really suggesting syntax, just contents.)

Which is getting quite hairy for something that is just meant to 
tide us over... But it shoudn't be that bad to implement, and
I don't see any other way of dealing with the "legacy font in
random encoding" problem. 

Regards,
                                        Owen