Re: Gnus reply encoding

[Stefan]

> I imagine the problem is that there are `small' encodings (like 8859*),
> and `big' encodings (like iso-2022-jp, utf-8, etc), and what you really
> want to say is "prefer the japanese big encodings to other big encodings"
> (whereas probably the default should probably be "prefer utf-8 to other
> big encodings").  For a message that's entirely with a small encoding you
> probably want to use the defaults.

Agreed.  We can think of it as an optimization: prefer an efficient single
byte encoding over any multibyte encoding.  The situation is different from
what you want with files since email is always tagged so you don't need to
worry about auto-detection when the email is read.

Regarding "small" encodings: I've recently received french email encoded in
windows-1256 (supposedly an arabic encoding), because those email happened
to use a set of chars that was present not just in latin-1 but also in
windows-1256 (typically: é and à).  Those email were not generated by Emacs
(our honor is safe here ;-), but it points out some potential risk we should
be aware of if we go too far in the direction of "use whichever encoding is
the most efficient without worrying about the locale".


        Stefan