[luatex] hyphenating ancient Greek

[Khaled Hosny]

On Sat, Jun 12, 2010 at 10:55:56PM +0200, Reinhard Kotucha wrote:
> On 12 June 2010 Khaled Hosny wrote:
> 
>  > On Sat, Jun 12, 2010 at 07:16:48PM +0100, Robin Fairbairns wrote:
>  > > Patrick Gundlach <patrick at gundla.ch> wrote:
>  > > 
>  > > > ... to set lccode for every greek letter (as far as understand it, but someone else should confirm).
>  > > > 
>  > > > Underfull \hbox (badness 10000) in paragraph at lines 29--29
>  > > > [][] \EU2/cmr/m/n/12 χα-λε-πὰ
>  > > > 
>  > > > ... whishing that this could be done at the engine level....
>  > > 
>  > > what?  setting uc/lccode tables?  that would require a further primitive
>  > > to specify the language (and probably also writing scheme -- i'm not
>  > > enough of a linguist to know), since the meanings and relations of
>  > > glyphs differ in some languages.
>  > > 
>  > > type example:
>  > > 
>  > > in english, the uccode of i gives you I
>  > > in turkish, the uccode of i gives you \.I
>  > >        (and the lccode of I gives you \i)
>  > 
>  > Unicode have uppercase/lowe case matching some where in its data, so we
>  > could just follow whatever unicode suggests.
> 
> Sure, but I fear that Robin's objection is still valid.  Unicode
> treats Turkish as an exception too, see:
> 
>    http://unicode.org/Public/UNIDATA/CaseFolding.txt
> 
> As far as I understand you have to know in advance whether you are
> typesetting Turkish.  A static table for all languages isn't
> sufficient.
> 
> It would be much more convenient if case folding wouldn't depend on
> the language, i.e. the Turkish "i" had a separate code point.
> 
> However, I think it's solvable.

Well, tex hyphenation still kind of black art to me, but if the tables
are to be language specific, then we shouldn't it in the format, but
rather hook some (lua?) code to the language activation sytem that would
do that on the fly, right?

-- 
 Khaled Hosny
 Arabic localiser and member of Arabeyes.org team
 Free font developer