[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#476957: (fwd) Bug#476957: texlive-xetex: Sinhala language support



Hi Jonathan & Norbet,

On Mon, 2008-04-21 at 20:46 +0100, Jonathan Kew wrote:

> XeTeX uses a version of ICU with some API extensions (to implement  
> features that ICU itself does not yet offer), so it has to use a  
> private build of the library.
> 
> Obviously, I hope that eventually ICU will support everything XeTeX  
> needs, and then we can change all this.

Fingers crossed!

> As part of TeX Live 2008. Current XeTeX trunk uses ICU 3.8.1 plus  
> some more recent patches from ICU svn, plus the XeTeX extensions.

Is there an estimated release date for TeX Live 2008?

> But TL 2007, which Norbert is currently packaging for Lenny, has an  
> older version of XeTeX that uses ICU 3.4 with extensions, hence this  
> issue.

I understand, however, if Tex Live 2008 is released before Lenny freezes
and can be packaged in time, then there's no need to go through this
additional package patching and respin. The Tex Live 2008 release date
and Norbet's packaging plans would determine if this is possible.

> I haven't seen patches to 3.4 in ICU's svn; is there a patched 3.4  
> branch or something that includes these? A full update from ICU 3.4  
> to 3,8.1 is too major of a change to slip into the TL2007 package for  
> Debian, I think.

The patches aren't *explicitly* for ICU 3.4. IIRC most of them were
applied to the pre-3.6 codebase. Here are Eric's ICU checkins which are
relevant to adding Sinhala support:
http://bugs.icu-project.org/trac/changeset/18370
http://bugs.icu-project.org/trac/changeset/18491
http://bugs.icu-project.org/trac/changeset/19979
http://bugs.icu-project.org/trac/changeset/19159
http://bugs.icu-project.org/trac/changeset/20656

Finally, to get correct Sinhala rendering you will also have to checkin
this fix (wait till Eric checks it into ICU's SVN):
http://bugs.icu-project.org/trac/ticket/6232


The issues with the submitted patch are:
1) The ooffice crashing workaround code, which renders incorrectly for
one codepoint, is made redundant by the increase of the worst case
expansion from 3 to 4.
2) The change to LEFontInstance.cpp does directly affect other scripts.
Probably more risky than the state table changes.
3) It still contains the bug mentioned in ticket 6232.

cya,
#




Reply to: