RE: Performance issue with new gtk 2.0 for win32

[email protected] writes:
 > 10 TIMES BETTER!!!! Thanx Tor, this makes the world of difference
 > on a p2 333 box.

That's odd, as I have only a 450 MHz P3 myself, and didn't notice that
dramatic a difference with Uniscribe or not. But I didn't check with
any "real" application that draws lots of text.

(For those who are unaware, Uniscribe is needed to render "complex"
scripts, like Arabic (not because of the RTL direction, but because of
the extensive ligatures that *must* be used) and Devanagari (*really*
different from Latin script). Chinese and Japanese are *not*
considered complex scripts, BTW, AFAIK, they just have lots of
characters.)

I probably should add code to basic-win32.c to bypass most of the
Uniscribe machinery for non-complex scripts. And do some benchmarking
and see where the time actually is spent, and check that I am not
using Uniscribe in some funny way that makes it slower than necessary.

--tml