[tex-hyphen] tamil and malayalam

Mojca Miklavec mojca.miklavec.lists at gmail.com
Fri Mar 19 19:23:01 CET 2010


On Tue, Mar 16, 2010 at 22:17, Karl Berry wrote:
>    TeX distributions like MikTeX
>    and TeX Live can decide whether the patterns for non-Western languages
>    that are only usable with xetex or luatex should be commented out by
>    default or not.
>
> I wish we weren't going down this road of non-universally-usable
> patterns at all, but maybe there's no good alternative.
>
> Anyway, in general, the TL developers like me can't possibly guess
> whether pattern foobar is only usable with xetex/luatex or not,

We have the exact information about whether a pattern is usable in an
engine or not (for example: ibycus is not usable in XeTeX, sanskrit is
not usable in pdfTeX). On the other hand this would enable using both
"mongolian:LMC" and "mongolian:T2A" (or the same for any other
language) that should be able to automatically use patterns with the
right encoding depending on the argument of fontenc. There was a
discussion about that in 2008: it seems that the author of Babel added
support for loading different encodings at about that time.

(We could do that now, but it would be a bit silly to load 10 Russian
patterns in "different encodings" for XeTeX.)

> So let's figure out a way to mark the non-pdftex patterns so that we
> (TL) have the necessary information to create a per-engine language.dat
> if we go that way.  Putting them in a separate directory is what comes
> to my mind.

It's not necessary to put them in a separate directory. We already
have that information. (When auto-generating loadhyph-xx.tex I need to
know for which languages I should not load anything for 8-bit engine.)
If we decide to introduce engine-specific language.dat, the easiest
way would be to extend the "addhyphen" directive in TL package
descriptions. (Maybe TL 2010 would offer better timing since it would
give us a bit more time to test.)

One question: if language.dat says
    somelanguage:T1 loadhyph-xx.tex
is it possible to access language name (somelanguage:T1) inside
loadhyph-xx.tex somehow? (This is not really needed; it would only
optimize the number of total files if we go that way.)

Summary:
- it would be nice to have engine-specific language.dat for the
reasons to follow
- XeTeX doesn't need ibycus, mongolian:LMC, mongolian:T2A, slovak:IL2,
slovak:T1, ...
- pdfTeX doesn't need exotic scripts (doesn't know how to handle them)
- if any of changes are to be made, I would time them for "TL 2010",
leaving a bit more time to do the testing
- a group is trying to improve loading of patterns for LuaTeX; we'll
take over whatever solution they come up with
- we have added 12 new patterns to the package (new version will be
released in a few days - we are waiting for the second response from
indic patterns maintaner)

Mojca

PS: Karl, maybe the "old and new" part from description of mailing
list may be removed now :) We are not dealing with old patterns any
more ;-)



More information about the tex-hyphen mailing list