Re: Emacs 22 branch created.

[Richard Stallman]

    >   * At the moment, it is slightly worse than current Rmail when it comes
    >     to non-ascii support.  One of the major points of moving to the mbox
    >     format is to be compatible with external tools, and that rules out
    >     storing mailboxes in emacs-mule form, which old Rmail does.  I'm
    >     still not sure what the best approach here would be here.  UTF-8?
    >     Raw text?

    The mailbox should not be decoded, most likely.
    The decoding should only happen when displaying a message.

That would mean decoding each message when you move to it,
and reencoding it when you move away from it.  And I presume
it would have to reencode the current message to save it.
Not good!

Or it would mean decoding it in a separate buffer, which has
its own big inconveniences.

I think it should decode a message when it is first visited (and the
extra header that Rmail saves data in is constructed).

(I don't recall whether I implemented that already in Rmail-inbox.)

I think the issue of decoding needs to be dealt with before
the Rmail-inbox branch is merged in.