On 2000.09.23, in <[EMAIL PROTECTED]>,
        "Jens Askengren" <[EMAIL PROTECTED]> wrote:
> 
> A GUI-mutt could be implemented by separating mutt into a backend and
> several frontends (curses, X11, etc). The frontend could be selected at
> compiletime, or loaded as a plugin/dll/.so-lib at runtime. Given the same
> .muttrc, the different frontends should feel and act the same.

A few years ago, I was tired of my frustrations with MH, elm, and
Columbia mm, and I hated Pine.  Although I didn't really want to use a
GUI, I had occasional uses for one, and knew others like having them,
too.

I had an idea then of producing a mailer founded on a small core, with
modules to provide the front-ends, the mailbox interfaces, and
script-language bindings.  One DSO for an X frontend, one DSO for a
curses frontend, one for an MH-like stateless system.  One for CGI or
mod_perl, maybe.  A DSO for IMAP, one for MH, one for mbox, one for
NNTP.  A DSO for perl, a DSO for python.  I wanted to provide
interfaces suitable to implement all these, in such a ways that they
could be used concurrently (although some UI combinations would be hard
to make sense of).

I was in the planning stages, and had written a tiny amount of core
code, when I found mutt.  I've been happily procrastinating ever
since.  (I don't really have that kind of time, anyway.)  Mutt is an
excellent mailer, good enough to resolve almost all my issues with the
mailers I've used over the years.  But I still like my original plan,
and I still want to see it done, just because I feel that it's the
right approach to a mailer's design.  In some ways, it's easiest now if
it's done on Mutt, but that's a LOT of code-mashing and munging.  I
don't know whether it's worth it anymore.  But I think that kind of
design is a REALLY good idea.  And, in that light, there's absolutely
no reason to bar a GUI front-end, even though I personally don't have
much use for one.

> Additionally, a GUI-mutt could have a frame based addressbook, a
> .muttrc-wizzard, etc.

Yes, but another interface could too.  That ought not to be dependent
on a GUI, although it might be most comfortable in a GUI.  That ought
to be an available interface to any willing UI.

> What do you think? If such an idea has a chance to go into the mutt-sources,
> I would like to contribute to a GTK-frontend.

I guess you know what I think now. :)  Yes, yes, yes, but I don't
really know whether it's feasible, and I certainly don't know whether
the mutt group is interested in this kind of fundamental change.

-- 
 -D.    [EMAIL PROTECTED]        NSIT    University of Chicago

PGP signature

Reply via email to