problem asserts in nnimap

problem asserts in nnimap

[joakim]

I'm connecting to a dovecot instance with nnimap.

I get problem with asserts failing.

I tried debugging it a bit like this:

in imap-parse-body in imap.el
...
	(when (eq (char-after) ?\ ) ;; body-ext-1part:
	  (imap-forward)
	  (push (imap-parse-nstring) body) ;; body-fld-md5
	  (setq body (append (imap-parse-body-ext) body))) ;; body-ext-1part..

(my-assert "1")
	(imap-forward)
	(nreverse body)))))
...
(defun my-assert(s)
  (assert (eq (char-after) ?\)) nil (concat "In imap-parse-body JAVE "
	s " " (buffer-substring (- (point) 50) (+ 1(point))))9
)
...
(i just factored out the assert)

in the backtace buffer i get this:
  signal(error ("In imap-parse-body JAVE 1 ted\")(\"text\" \"x-vcal\" (\"charset\" \"iso-8859-1\") {25}"))

I interpret this as meaning that an ")" is expected, but instead we
got "}". 

in *imap-log* it looks like this:

("text" "x-vcal" ("charset" "iso-8859-1") {25}<startupmöte_forts._.ics> NIL "quoted-printable" 550 22)


Im not too familiar with gnus inner workings or nnimap, so any
assistance would be apreciated!

-- 
Joakim Verona
http://www.verona.se

Re: problem asserts in nnimap

[gdt]

I am also using dovecot with nnimap.  I get failures with
imap-parse-body but it continues (without debug-on-error set).  This
seems to happen with only some messages and I haven't figured out
which one.

-- 
	Greg Troxel <gdt@work.lexort.com>

Re: problem asserts in nnimap

[joakim]

gdt@work.lexort.com writes:

> I am also using dovecot with nnimap.  I get failures with
> imap-parse-body but it continues (without debug-on-error set).  This
> seems to happen with only some messages and I haven't figured out
> which one.

This seems to be exactly the problem I'm experiencing.  If I dont set
have debug-on-error set, gnus will carry on and ignore the error, but
the message will not be handled correctly.


> -- 
> 	Greg Troxel <gdt@work.lexort.com>
>
>

-- 
Joakim Verona
http://www.verona.se

Re: problem asserts in nnimap

[gdt]

I've been losing with "imap-parse-body 2" for a while, and believe
I've tracked down what sorts of messages cause it.

Basically, entering a group with g, or rescanning with M-g causes an
assert to fail in imap-parse-body.  Messages which have a
message/rfc822 part seem to all trigger the problem when in INBOX, and
I searched for the messages by using 100 M-G and then 102 M-g to
binary search for most recent problem message.  I then used
Thunderbird to move messages to a "Trouble" IMAP folder, since Gnus
won't put them in the summary.  One prevented even getting a summary
buffer.  Messages with other parts, such as attached pdf, and PGP/MIME
seems fine.

I can look at the Trouble folder just fine with gnus.  But the same
messages in INBOX hit the assert.

It would be nice if the assert included the UID, but I don't see how
to do that.

I'm using

  NetBSD/i386 -current
  emacs 21.4
  up-to-date CVS Gnus
  dovecot

With the same IMAP server, I have no issues when using Thunderbird
(other than that my wrists will hurt from using the mouse).

-- 
	Greg Troxel <gdt@work.lexort.com>