Partially implemented feature.

[visigoth@naiad.fac.cs.cmu.edu]

I've written code that extracts the base UID for an IMAP group from
the Group parameters.  This way I can get at the info before I'm given
the group-info to mung.  I figured that the race between when I use
that information and when I get to mung (and at that point store) the
information is something I could deal with for now.

Unfortunately, when I wrote my munging code, I noticed that absolutely
nothing is happening to the information returned from
nnimap-request-update-info.  The documentation describes it as being a
filter.  In any case, this is a problem, and there are several layers
of stuff between when the info gets passed into the guts of Gnus and
when it gets to my routines, and then comes back again, and I don't
trust my knowledge of the internals well enough to muck with that
right now.

I've made my code adjust the group-info destructively for now, and I
don't know how safe that is.  But it is, at the least, working quite
well for me.

If I could only figure out a way to get 11000 group-lines out of an
IMAP server without taking five minutes, I'd be happy.

I'm considering making available the default behavior for nnimap to
snarf down information only about groups that are subscribed on the
server.  For a big server like ours, it's damnably useful.  I can also
provide tools to let you sub and unsub from inside emacs, but I'm not
sure about inside Gnus.

Another possibility is to make multiple virtual servers--a virtual
server can describe what subset of the group on the server it wants
(i.e. "alt.sex.*" would be a pattern for matching certain groups.)
Then each server only pulls that subset in.  If you have no interest,
say, in the local collection of vms.* stuff, you don't include it in
your patterns.

I'm almost ready to go public with this.  Any comments?

(It supports, just so you know, IMAP4rev1--not IMAP anything older.  I
may add support for older stuff when I get a chance.)

John.