Re: hitex format 2026

On Sun, 8 Feb 2026 at 22:00, Karl Berry <[email protected]> wrote:

> David, Ulrike, all - thanks for the reports.  What to do is up to
> Martin. I already reported the failures in hitex/hilatex fmt building to
> him and disabled the building in the sources (hitex.tlpsrc).
>
> So, Ulrike, I don't know why any hitex-related fmt was still trying to
> be built for you. If that's what was happening.


Ulrike had a local fmt configuration that caused it to build (or rather not
build)


> If hitex.fmt is going to
> stay disabled for a while, I'll update the news.
>
> David, it is not simple to reduce the hyphenation files being
> read. Right now hitex is sharing the same plain-related hyph setup as
> all the other engines, which loads all the hyph patterns. I am not
> enthused about changing that. Martin said the problem is not just how
> many are loaded, but that the "Church Slavonic" hyph-cu.tex hyphenation
> file (and maybe others, didn't go beyond that) uses Unicode chars which
> are intentionally not supported. --thanks, karl.
>

Yes if you comment out Church Slavonic, then later hyphenation files fail.
(I wanted to build a plain format for testing so I stuck \endinput half way
down
language.def, but I realise that's not an option for a distributed
solution...)



This is really a similar issue to the latex failure, the utf-8 handling
means that you can't
use the classic 8-bit code paths. but the Unicode support isn't enough to
handle Unicode
in any reasonably small update to the xetex or luatex code paths.

David