Re: sh command bogus output in UTF-8 text

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Fri, 7 Feb 2020 05:02:13 -0000 (UTC)
    From:        mlelstv%serpens.de@localhost (Michael van Elst)
    Message-ID:  <r1ir0l$frl$1%serpens.de@localhost>

  | sh in netbsd-8 seems to be fine.

It was for this particular usage, but gets those chars incorrect in
other places.   In general sh is much more clean now than it was, just
here docs with unquoted end delims (ones that are internally processed)
got missed from the final processing somehow.

I have a working version (different patch entirely from the one I sent
to the list earlier) but I'm not entirely happy with the way that it is
implemented, so I am looking for how to make it better before committing
it.

  | The shell isn't really 8-bit clean, internally the parser uses control
  | characters with byte values 129..139 that are supposed to be escaped
  | on input.

They all are now, there's no longer any "supposed" (or none that I have
been able to find) ... the problem here is in a failure to undo the
escaping before output.

kre