MH-E folder completion

[Bill Wohler]

Hi Stefan,

I'm *finally* getting around to testing your completion code that you
added to MH-E. Your original outline was:

    1- use completion-in-region (new in Emacs-23.2) instead of
       mh-complete-word 
    2- change the completion tables to obey the completion-boundaries,
       which should also make them support partial-completion (e.g.
       complete +a/b/c to +alt/bob/calls).

The second feature sounds very useful, and it even seems to be working!
Which is amazing because you didn't have MH with which to test.

I'll be testing your fallback code on XEmacs21 momentarily, but first I
share to share one observation and discuss it. The old behavior looked
like this:

    Visit folder (default +inbox): +lists/TAB

    Possible completions are:
    +lists/acm          +lists/alerts/
    +lists/all-battery  +lists/aob
    +lists/appframework +lists/aquadivelog
    ...

However, the new behavior looks like this:

    Visit folder (default +inbox): +lists/TAB

    Possible completions are:
    acm         alerts/         all-battery
    aob         appframework    aquadivelog
    ...

I think the old behavior looks more natural to an MH user since you're
building up a folder string that you could pass to an MH command.

I'm guessing that the Emacs folks would say that the latter behavior is
better for Emacs users, and that the complete folder string can be
viewed in the minibuffer. Yes?

I'd like to open up the discussion to Emacs and MH-E developers to
ensure that the new behavior is preferable for MH users and Emacs users
alike, and if not, what the appropriate behavior should be. Thanks.

-- 
Bill Wohler <address@hidden> aka <address@hidden>
http://www.newt.com/wohler/
GnuPG ID:610BD9AD