>>>> "Stephen" == Stephen J Turnbull
>>>> "Jan" == Jan Rychter <jan(a)rychter.com> writes:
Jan> And BTW, I do have (setq default-buffer-file-coding-system
Jan> 'iso-8859-2), I'm surprised I got iso-2022 output at all, but I
Jan> guess I missed something here.
Stephen> The use of ISO 2022 in ISO 8859 coding systems is something
Stephen> the Japanese insisted on in traditional Mule, and is simply a
Stephen> violation of the standard (in spirit, if not the letter).
Stephen> I'll do something about it soon, but it's a major,
Stephen> backward-incompatible, change. Done wrong it could result in
Stephen> unrecoverable corruption. As it stands the corruption is
Using GNU Emacs, that is (for me) -- I've been unable to recover the
file using XEmacs.
Stephen> Try the latin-unity package for your main problem. There's a
Er, well, I remember I have tried latin-unity a while ago and it worked,
but right now all I get when I call any of its recode functions is:
Wrong type argument: stringp, t
That happens also if I start xemacs with -q.
But -- that leaves my main question open -- is the following sequence
supposed to work in XEmacs and produce an unibyte file, just as it does
in GNU Emacs, or not?
-- open the file with iso-2022 characters
-- set the coding system to the one you know can represent all
characters in the buffer
-- write the buffer to a file