Re: 0.126, xemacs, pgnus-0.95, and really large UIDs

[Simon Josefsson <jas@pdc.kth.se>]

Michael Poole <poole@graviton.subatomic.org> writes:

> I could, yes.  It does not seem to me that it should be so hard to
> have an elisp variable to toggle between LIST and LSUB as the
> default select method.

IMHO the proper solution is to add server-side subscribedness
awareness to Gnus. Waiting for that to happen, I modified your patch a
little and applied it, thanks!

> Thus (the diff would be half as long, but it looks like
> imap-mailbox-list got extra functionality at some point but
> imap-mailbox-lsub did not get a corresponding update):

No, that was intentional. `imap-mailbox-list' need extra functionality
to limit number of mailboxes to list, `imap-mailbox-lsub' do not need
such functionality since you can assume the number of subscribed
mailboxes is quite low.

OTOH a consistent interface is nice, so I added your patch, thanks!

> There are a few other places where UIDs are assumed to be stored as integers.
> Even once I do fix the places in nnimap where things fail typechecks, things
> don't quite work for some reason.

Yes, Gnus uses integers for article numbers.

If you want to debug it further, I recomend looking at what nnimap
puts into `nntp-server-buffer', that need to conform to the
specification, and any integers there will be read with `read' so UIDs
need to be mapped to integers (<2^28) before ending up in the server
buffer.

Having a bignum integer type in emacs is something I'd offer lots of
beer for... :-)