What about implementing a solution with luatex? Would that work?
No. I mean, sure, it could work for people using luatex, but
a) luatex will be a moving target for quite some time to come, and
b) a vanishingly small percentage of people use it, and
c) it would take quite a lot of development time to figure out what to
do and then do it. Processing through m4, OTOH, should be fine.
Sounds reasonable. However, if you do that, m4 *must* become part of
TeXLive!
Gee. I totally disagree! It is not TL's business to provide system
utilities. The resulting conflicts are a true pain. Even getting it to
build on our all platforms, in a way that works in the builders'
environments, would not be simple. I'd remove texinfo.tex from TL
before I added m4 ...
> \input texinfo @c -*-texinfo-*- `m4needed'
Hmm. Why not also using Emacs file variable syntax for indicating m4?
Simplicity/robustness of parsing? I just want to grep for something to
determine if it should be passed through m4.
^\\input texinfo.*@c.*`m4needed'
People can do a lot of tricky stuff in Local Variable sections. Heck,
just finding them is nontrivial. Anything can be done, but ... the
simpler the better.
k
