Mule bugs: misidentification (Latin-1 vs. Chinese), revert issues

[stephen at xemacs.org]

Michael Sperber writes:

 > I've running it by default for a few months now, and whenever
 > anything connected to encodings comes up, XEmacs screws up royally.
 > (Like identifying 8859-1 as Chinese, losing encoding information on
 > revert, etc.)

8859-1 == Chinese should be fixed now.  There will be more of these
lurking; code identification is a neural-network kind of thing
(unfortunately without the learning part at this point).

If you get a misidentification, send a bug report with the expected
and actual encodings, the file (if you can legally/ethically do so),
and the value of LANG when you started XEmacs.  (There are internal
variables that I should add to M-x report-xemacs-bug, but they're all
defaulted from LANG, and it's a pretty safe bet you don't mess with
them directly.)

Losing encoding information on revert?  Please be more specific.