On Fri, Jul 14, 2006 at 11:15:23PM +0430, Hossein Noorikhah wrote: > Hi. > I'm using Lars Gullik Bjønnes's Unicode branch of LyX 1.5, and I'm > interested in writing Arabic/Persian documents with it. I want to know > the status of this branch, and also how can I find more information > about it. > To me, It seems that the original bidirectional algorithm that is used > for the old Arabic/Hebrew patches doesn't work here.
It is a kludge. I am somewhat familiar with it as I tried to fix it in 1.4 (in the blind, as I don't know any RtL language). > I've compiled, and tested this branch today, and It seems that in > English mode, character shaping for Arabic goes well, but it does not > apply bidirectional algorithm. When I enable bidirectional > option(Right-to-Left languages support) from Tools > menu/Preferences/Language and I set document language to Arabic in > Document menu/Setting/Language, everything goes wrong. You can look at > the attached image. I've attached both the wrong answer I get from > Unicode branch, and the right thing that is produced by the original > Arabic version (i.e. LyX 1.3.5) that uses iso8850-6 encoding. > I think the bidirectional code should be rewritten to match the > Unicode branch. May be we can use libfribidi(http://www.fribidi.org) > to handle bidirectional text. > Any ideas on this issue? I suspect the problem is due to our kludge and the new Unicode stuff both trying to do the right thing, and the net result is wrong... my guess. Obviously with Unicode enabled, the kludge should die. The problem as I see is that we do out own brute-force character positioning on output. > Thanks. - Martin
