Re: Emacs-diffs Digest, Vol 100, Issue 28

[Karl Berry]

What I can tell you is that texinfo.tex is currently limited to the
Computer Modern fonts and a few extras.  So most Latin-based languages
are more or less ok.  Anything using other scripts -- Cyrillic, Greek,
Indian, Chinese, Arabic, Urdu, whatever -- is not going to work.  You
might not get an error from TeX but you won't get any useful output
either.  I suggest checking the .log for "missing character" messages.

As far as makeinfo goes, all currently released versions blindly pass
the input to the output as far as such "binary" characters go.  So if
you write some Chinese UTF-8 in the input, those bytes will end up in
the output just as written.  No actual Unicode processing takes place,
as required by Hindi and Arabic, for example.

Although I am not one to downplay the importance of correctly
typesetting people's names, I question whether doing so requires using
UTF-8 for the manual.  After all, the manuals are about Emacs and are
written in US English using 7-bit ASCII, not about stressing the
typesetting system by expecting to be able to handle every script in the
world in one document simultaneously.

Here's a kludge that comes to mind as a possible workaround: Perhaps the
names could be written in a separate document that uses standard LaTeX
(or XeLaTeX), where all the necessary fonts are at least conceivably
available (though don't expect it to be zero effort to get it all to
work).  Or perhaps it takes several small documents, one for each major
script family.  The separate documents could be included in the manual
in their .pdf form for the TeX output.  

Best,
karl