On Tue, Aug 29, 2006 at 02:24:06PM +0200, Peter Kümmel wrote: > Abdelrazak Younes wrote: > > Andre Poenitz wrote: > >> On Tue, Aug 22, 2006 at 10:44:21AM +0200, Abdelrazak Younes wrote: > >>> I proposed to use ucs2 conditionally at compile time to avoid the > >>> above conversion but Lars is furiously against that. I really don't > >>> know how we are going to represent an ucs4 characters on screen with > >>> Qt4 so I don't see the reason to use that. Or is there a way to do that? > >> > >> UCS-4 -> UCS-2 conversion should not be noticable compared to displaying > >> the result on screen. > > > > I thought also that it should be a matter of a simple cast from 4 to 2 > > bytes, thus ignoring the trailing two bytes (or the first two for big > > endian systems). Actually I have been experimenting with that and it > > appears that it works well. > > > > In this patch I use simple casts instead of the costly ucs2_to_ucs4 and > > ucs4_to_ucs2 functions. It turned out that all display issues are > > resolved... including math! It even brings a very nice speed-up... > > > > Try it :-) > > > > I've tested it under Windows and Linux and the patch fixes all the > display issues, great.
I cannot get some math symbols right, most notably greek characters. -- Enrico
