Re: [PATCHes] Add basic multibyte charset handling to makeinfo

[Karl Berry]

    I don't think it is really practical to make
    texinfo dependent on some other library

Yeah, worse than the disease for sure.

    The standalone info reader ignores the "Local Variables: coding: ..."

The Emacs reader does pay attention to that (I believe).  At least,
that's why the Local Variables: stuff got added in the first place,
cause Emacs wanted it.  I'm not sure how or if encoding name differences
get handled.

    Thus I prefer a model in which all info files installed on the system
    use a common character set

It's a nice theory, but realistically, it won't happen.  People will
write (have written) documents using

@documentencoding ISO-8859-1
and
@documentencoding ISO-8859-2
etc. and I sincerely doubt any Info documents have been "converted by a
distributor" to UTF-8, ever.


However, I see no real solution either, and I doubt the discrepancy will
make much difference in practice -- there are few if any Texinfo
documents written in Asian languages, and I think the new code will have
basically no effect on Latin-based languages?.  I think your work is an
improvement over what we have now for sure, and it's not obvious how to
go further.  So fine.


This reminds me of the related issue of separating per-document
translations (like "Chapter", which should be based on
@documentlanguage) from per-program messages (like the --help output).
This, unlike the encoding thing, should be doable, since gettext can
switch between locales dynamically (there is some examples in the
gettext distribution).  I got about halfway through implementing the
change, with help from Norbert Preining, and then we both got lost in
other projects.

Since you are delving so deeply into makeinfo already, maybe there's a
chance you could look at that?  Finishing that work is really the
biggest thing on my list before doing another release.

Thanks,
Karl