[luatex] Hyphenation in plain TeX

[Vladimir Lomov]

** Mojca Miklavec [2010-10-08 11:21:32 +0200]:

> On Fri, Oct 8, 2010 at 11:04, Robin Fairbairns wrote:
>> Mojca Miklavec wrote:
>>
>>> This should be fixed in format one day though, but if you want to use
>>> it now, this is what you need to do ...
>>
>> what should be fixed in the format?  lccode tables for all unicode
>> ranges covering alphabetic characters?
> 
> Exactly.
> 
Won't argue. I think how it is not big problem have all them in format.
May be 10-20 years ago this could be a problem but now I doubt.

>> seems extreme:
> 
> Well, it's done by:
> - plain XeTeX & XeLaTeX
>   /usr/local/texlive/2010/texmf-dist/tex/generic/xetexconfig/unicode-letters.tex
> - ConTeXt when using XeTeX
>   /usr/local/texlive/2010/texmf-dist/tex/context/base/xetx-utf.mkii
> - ConTeXt MKIV
>   /usr/local/texlive/2010/texmf-dist/tex/context/base/char-def.lua
> 
> I don't know why it would be any more extreme to do it in plain LuaTeX format.
Thanks. I looked at them. But could luatex use the same
unicode-letters.tex? if could then this makes luatex to work with
accented letters as well as cyrillic (and other?).

>> i would think it
>> more sensible to provide the tables in hyphenation files
> 
> Btw, there is
>     /usr/local/texlive/2010/texmf-dist/tex/generic/hyph-utf8/patterns/txt/hyph-ru.chr.txt
> but it is not used anywhere.
But are other files (like hyph-de-1996.chr.txt or hyph-fr.chr.txt) used?

>> that way
>> there would be some chance that the tables were constributed by people
>> who know the language.
> 
> Now imagine user A using both German and French patterns. When he
> switches from German to French - should we switch all the German
> lccodes for äöü back to undefined? If we don't, there's a chance that
> the user will get different hyphenation for French (let's say that the
> text contains a name with some German letters) depending on whether he
> was typesetting a French-only or mixed language document.
> 
> Now imagine user B using some package that deals with Unicode and sets
> all the lccodes for the whole Unicode range. If we do change the
> lccodes back to undefined when switching from German to French, the
> user will have all the letters for the whole Unicode range properly
> defined, except for some umlauts which we modified without user's
> consent.
> 
> I simply don't think that changing any codes inside hyphenation files
> is the right thing to do. Yes, we can solve many users' problems
> (simply hyphenation will work), but create many ugly new ones.
Don't get the idea. As ordinary user (and wokring with more ordinary
users) I think that it is best if all would work (even if format file is
big).

Consider the other reason: now latex format file in TeX Live has
patterns for many languages (latex simple.tex and see the first few
lines) but I don't see that many users try to reduce the number of
patterns enabled by default.

>From my point of view (user): the luatex is unicode aware engine
(actually I think of UTF-8) and this is benefit for me. From that point
of view having lccode's for (almost all) Unicode letters (European
languages, for example, Cyrillic, Greek, other?) may be expect Eastern
languages (because I'm not aware about they specific in compare with
European languages, this is another problem) it is natural.

---
WBR, Vladimir Lomov

-- 
<barneyfu> knghtbrd: crap, SDL sure makes DGA a helluva alot easier too
           doesn't it? :)
<knghtbrd> barneyfu: what DGA?
<barneyfu> mouse dga
<knghtbrd> barneyfu: (does that answer your question?)
<barneyfu> Hahahahaha YEAH! :)