Summary functions damage vm-v5-data headers

[Uday Reddy]

Johan Vromans reports (viewmail-info, 2011-08-17):

Using the attached mailbox (small) I can reproduce the following
problem:

* Start GNU/Emacs 23.2 (Fedora14) with "-q" command line option.
* Load VM 8.2.0a (vm-8.0.2a/vm-load.el).
* Mx vm-visit-folder, and point it to the mailbox.

4 messages, all new, are shown. The summary shows "Sheila en René".

* Visit all 4 messages; the status changes from New to Read.
* Save the mailbox.
* Exit Emacs

* Restart emacs.
* Load VM 8.2.0a (vm-8.0.2a/vm-load.el).
* Mx vm-visit-folder, and point it to the mailbox.

4 messages are shown. The summary shows "Sheila en Ren\351".

Inspection of the (saved) mailbox shows that in the X-VM-v5-Data header,
the name "Sheila en René" is stored encoded in Latin1. I very much doubt
if that's correct.

[Uday Reddy (reddyuday)]

Follow-up

So it must be some setting in my preferences.

Hmm....

My summary format is:

  (setq vm-summary-format "%n %*%a %-18.18UB %-3.3m %2d %4l %UL%I%s\n")

The actual entry in the summary is generated by vm-summary-function-B,
which is part of the bbdb package (file bbdb-vm.el).

  (defun vm-summary-function-B (m &optional to-p)
    "Given a VM message returns the BBDB name of the sender.
  Respects vm-summary-uninteresting-senders."
    (if (and vm-summary-uninteresting-senders (not to-p))
        (let ((case-fold-search nil))
      (if (string-match vm-summary-uninteresting-senders (vm-su-from m))
          (concat vm-summary-uninteresting-senders-arrow
              (vm-summary-function-B m t))
        (or (bbdb/vm-alternate-full-name (vm-su-from m))
            (vm-su-full-name m))))
      (or (bbdb/vm-alternate-full-name (if to-p (vm-su-to m) (vm-su-from m)))
      (vm-decode-mime-encoded-words-in-string
       (if to-p (vm-su-to-names m) (vm-su-full-name m))))))

This looks fine at first sight, though.

Uday's response:

No, it isn't working. As soon as I tried to use it, I got bad characters in
the summary. BBDB hasn't been maintained for a long time, and trying to
puts its data into VM headers isn't very wise.

I will put this on the To-do list to find some decent solutions.