[texdoc] Weird behavior of `texdoc'

[Yudai NAKATA]

Dear all,

Sorry for late response and thanks for Karl & Paulo’s thorough investigation.

As I don’t have Lua installed on my ends, I tested only with texlua
and could reproduce
the issue without invoking texdoc.

> But we've never received a single
> report before this. It seems ... unlikely.

In my opinion, one possible reason for this is that it’s rather
unlikely for one to happen to see
a plain-text documentation containing non-ASCII characters via texdoc.
That being said,
there are many many TeX Live users in the world, so it’s still unlikely...

> I don't suppose it could be about that particular UTF-8 char?

In bxjalipsum example, all the Japanese characters turned into the
<XX> sequences and
eszett ß also corrupted in my quick test. So this doesn’t seem to be a
problem about specific
characters.

Regards,
Yudai Nakata

On 2017年2月16日 at 20:04:18, Paulo Roberto Massa Cereda
(cereda.paulo at gmail.com) wrote:
> LESSCHARSET=utf-8 texlua test.lua
>
> and it works. :) Looks like 'less' is the one to blame...
>
> Paulo
>
> Em 15-02-2017 20:31, Karl Berry escreveu:
> > but the underlying texlua interpreter.
> > One for dev-luatex at ntg.nl, perhaps? Karl, Manuel?
> >
> > I'm not sure yet.
> >
> > I reduced the input to a one-line file, call it /tmp/test.txt:
> > a → b
> >
> > And then running
> > env LANG=en_US.UTF-8 texlua /tmp/test.lua
> > with your test.lua, which calls less, gives the bad result.
> >
> > But if I change test.lua to the one line
> > os.execute('cat "/tmp/test.txt"') -- instead of less
> > the output is correct for me (given a UTF-8 terminal font, of course).
> >
> > Also, if it was a generic problem, every UTF-8 character would be
> > rendered incorrectly for everyone. But we've never received a single
> > report before this. It seems ... unlikely.
> >
> > I don't suppose it could be about that particular UTF-8 char? --karl
> >
>