Re: recent problem with IMAP counts

[Didier Verna <didier@xemacs.org>]

       I found the culprit. The culprit is:

didier(pts/0)% git log -1 -p -r 90c1f88e5357d61a732d6f78d3c9a0278d78f49e
commit 90c1f88e5357d61a732d6f78d3c9a0278d78f49e (refs/bisect/bad)
Author: Lars Magne Ingebrigtsen <larsi@quimbies.gnus.org>
Date:   Thu Jun 30 01:09:38 2011 +0200

    Bail out in nnimap if the server has closed the connection.

diff --git a/lisp/nnimap.el b/lisp/nnimap.el
index 2cfc889..e78c20b 100644
--- a/lisp/nnimap.el
+++ b/lisp/nnimap.el
@@ -1227,6 +1227,10 @@ textual parts.")
 
 (deffoo nnimap-finish-retrieve-group-infos (server infos sequences)
   (when (and sequences
+            ;; Check that the process is still alive.
+            (get-buffer-process (nnimap-buffer))
+            (memq (process-status (get-buffer-process (nnimap-buffer)))
+                  '(open run))
             (nnimap-possibly-change-group nil server))
     (with-current-buffer (nnimap-buffer)
       ;; Wait for the final data to trickle in.


But in fact, the problem is more complicated than I thought. I
eventually figured out that it's one of my nnimap servers that causes
the problem, and the problem exhibits different symptoms before and
after this revision of Gnus. I have two nnimap servers in Gnus (one on
gmail, the other somewhere else; I think it uses Zimbra).

Before this revision of Gnus, it sync'ed with gmail all right at
startup, and then it hung on the other server. The minibuffer displayed
"Retrieving ... [0%]" flashing and blinking all over the place. I had to
C-g on it and only then got my Group buffer. The gmail account had the
correct info, but not the other one.

Now, with the current HEAD, Gnus startup doesn't hang anymore, but none
of the servers get the correct information. If I remove the problematic
server from my secondary select methods, everything's fine again.

-- 
Resistance is futile. You will be jazzimilated.

Scientific site:   http://www.lrde.epita.fr/~didier
Music (Jazz) site: http://www.didierverna.com