Re: gnus and my new IMAP implementation

[Simon Josefsson]

Adam Megacz <adam@megacz.com> writes:

> Hi.  I'm debugging an IMAP implementation I wrote (part of the
> gmail-to-imap bridge), and I've run into some problems with GNUS'
> nnimap.  I'm almost certain that these are my fault, but they only
> happen in GNUS, so I figured I might be able to get some advice here
> on where to look.
>
> There are a couple of problems, but the only one I can consistently
> reproduce is GNUS' misestimate of the number of messages in a mailbox.
>
> The conversation below is the result of starting up gnus (fresh
> .newsrc.eld), selecting a mailbox ("newmail") containing 8343 messages
> (20 UNSEEN), pressing 'q' to return to the groups buffer, and then
> using C-u RETURN to view all messages (not just unseen) in the folder.
> Upon doing that, GNUS says
>
>   how many articles from newmail (default 13649):
>
> Where is 13649 coming from?  I can't figure this out ;)

It isn't coming from imap.el nor nnimap.el -- Gnus calculate the value
based on the knowledge it has.  Unfortunately, Gnus isn't smart enough
to ask for more information from the backend that would help it give
an accurate value.  So if your server has large holes in the UID
range, the algorithm Gnus uses for this will give bad results.

Pressing M-g on the group should make Gnus fetch more information,
though, but it is slow.

I guess the answer is that you will have to trace the computation of
the figure inside Gnus.  Naturally, Gnus uses information from the
backend during the computation, but I don't recall exactly how it
works.

I think this is a known problem, and it isn't specific to IMAP
although it normally occur more frequently with IMAP since servers
often have large UID holes.