[Devel] Re: [Fonts] Re: [Render] Font name matching (urg)

[Keith Packard]

> We have already discussed how to proceed (writing a text layout engine
> on top of FT2); some experimental (non-working) code how to parse OT
> tables can be found in the `ftlayout' CVS model; a replacement for
> glib is in the `mlib' module (which satifies the needs of FT2), etc.
> It will take some time...

Given the general lack of OT fonts, I suggest that we instead rely on Pango
(or other) libraries to perform text layout and leave my libraries at the 
glyph level.

One useful piece of information which I can't figure out is how to compute 
which unicode subsets a particular TrueType/Type1 font supply.  XFree86 
already has mapping files from 10646 to various national character sets; 
at some point I'd like to take advantage of those mappings and advertise 
the available encodings for each font, but I don't know how to discover 
which glyphs are available without enumerating the 10646 space.

> The only `high-level' stuff is synthesizing a Unicode character->glyph
> map.  Or do you mean something different.

I meant layout level stuff.  As FreeType 2 doesn't now, and probably won't 
any time soon support text layout, I don't think there's much conflict.  
We have:

        Pango           Layout, charset -> glyph conversions
        Xft             Font file/core font access, glyph presentation on the 
screen
        FreeType        Glyph rasterization

Xft will probably also support trivial character->glyph conversions for 
locales with simple 1<->1 mappings, as provided by the existing XFree86 
encoding tables.

address@hidden   XFree86 Core Team              SuSE, Inc.