Taco Hoekwater
David Kastrup wrote:
Well, the various LaTeX hyphenation pattern files are giving me a run for my money: authors think it is a good idea to encode them in binary, regardless of whether we are talking latin-1, T1 or koi8 encodings.
What a mess. I started writing to a few of the maintainers...
For simplicity, you should use the utf-8 encoded patterns used by xelatex and context. The trick with the callback makes luatex stop complaining, but it will not hyphenate correctly.
Why not? As long as one uses fonts in T1, T2A and similar encodings, things should work, shouldn't they? Of course, when using Unicode codepoints, this breaks down. But at the moment I am just aiming for getting the same behavior as with standard LaTeX.
XeTeX has run into the exact same problem.
Hardly surprising, yes.
The LaTeX way is (probably) creating a 'lualatex' format.
That's just for dumping: I have not actually tried running this format. I don't even know whether the callback survives into the format or not...
That functionality could be added in the future, but it doesn't survive (un)dumping at the moment.
Well, I guess that calls for \everyjob then at the moment. Or I should see what xelatex does. It is probably easier implementing the xelatex input encoding primitives in Lua than trying to think about all of this too much. -- David Kastrup