Re: Xft version 2 patches for gtk+ and pango

[Keith Packard <keithp keithp com>]

Around 10 o'clock on Feb 20, Owen Taylor wrote:

> Because this is in the marked-unstable portion, we can add to the API in
> 1.0.x and use it from the modules. Even in the marked-unstable portion
> though, I'd rather avoid removing API calls.

Then let's at least add the new calls.  I can make Xft internally track 
the number of outstanding locks for a face held by a font and release all 
of those when the font is closed; that will make it possible to emuate the 
'get' functionality with a call to 'XftLocKFace' when using the version-2 
API.

I was also wondering if we couldn't clean up the Unicode coverage stuff in
pango; currently, I'm copying the coverage information from Fontconfig into
the pango-specific structure, but that wastes some memory and a bit of
time.

Could we make the coverage checks abstract so that the different font 
drivers could manage them separately?  Fontconfig uses a sparse bitmap 
representation for coverage which is relatively efficient in both space 
and time, it also caches this information on disk to avoid recomputing it 
each time fonts are loaded.

I'd also like to find a way to use the cached coverage information to help 
pango locate fonts for otherwise unknown glyphs; right now, I get a lot of 
boxes on the screen.

Keith Packard        XFree86 Core Team        Compaq Cambridge Research Lab