[HarfBuzz] COMING SOON: 2nd Annual Text Layout Summit at aKademy in July 2007

[Kenichi Handa]

In article <1181657075.4331.24.camel at behdad>, Behdad Esfahbod <behdad at behdad.org> writes:

> Right, but per-process memory is more about malloc()s you do, and I
> guess parsing FLT data at run-time does a lot of those.  Reducing this
> per-process footprint is why I'm rewriting harfbuzz to use mmap()ed
> fonts directly instead of parsing them into memory.

FLT is parsed just once when loaded, and I think the memory
to keep the result of parsing is negligible (especially
compared with sizes of fonts).  Most FLT files are about 5K
bytes (usually more than half of it is a comment).

> > At first, I'd like to read the documentation of that
> > library.

> There's no documentation.  It was a big lump of code handed to us from
> freetype 1 and we are just maintaining it for now, and working on an
> improved version.

Ummmm, I hope at least the header files contain how to use
each API.

> > Could you tell me what CVS command to use?
> > 
> > % cvs -d ???? co ???/HarfBuzz

> Oops, we've moved to git now.  I don't have the details handy.  Simon,
> can you please update the wiki, pointing out which branch is stable and
> all...

Could you please also show an example command to check out
something like:

  % git ....

I've never used git.

---
Kenichi Handa
handa at m17n.org