Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

VM-8.0.12, Lotus Notes IMAP server and corrupted emails

0 views
Skip to first unread message

John Stoffel

unread,
Oct 27, 2009, 5:14:26 PM10/27/09
to
Hi guys,

Anyone else using VM with IMAP to a Lotus Notes server? I tried it
today, but I got a bunch of corrupted and mashed together emails, so I
had to back away quickly before I screwed everything up completely.
I'm back to Fetchmail instead, which stinks because now my BB isn't
kept in sync any more either.

Maybe I need to just leave all the mail on the server and not try to
pull them down at all? Dunno... but VM IMAP isn't very robust at all.

Thanks,
John

Uday S Reddy

unread,
Oct 27, 2009, 10:06:48 PM10/27/09
to
John Stoffel wrote:

> Maybe I need to just leave all the mail on the server and not try to
> pull them down at all? Dunno... but VM IMAP isn't very robust at all.

I disagree. People have been using VM with IMAP servers for a very long
time. IMAP is an internet standard and Kyle Jones did a careful job of
reading the spec and making sure that everything is done to the T. It
is more likely that the Lotus Notes is at fault. Being a proprietary
product, it can make its own rules which don't agree with the rest of
the world.

The error message you received clearly stated it as an "IMAP protocol
error" which means that there is a mismatch in the handling of the
protocol between the server and the VM. Your server isn't following the
IMAP protocol properly. We don't know what it is doing, but VM detected
a funny character which shouldn't have been there. It could be a minor
thing or it could be a serious fault.

You would be best advised to be cautious. Please check with your system
administrators what you can and cannot do with this "IMAP server", get
them to set up a test account for you to experiment with, and use a
variety of IMAP clients with this server to see which of them are able
to talk to it safely or if at all.

Cheers,
Uday

John Stoffel

unread,
Nov 12, 2009, 1:42:09 PM11/12/09
to

Hi Uday,

Thanks for the reply, and sorry for the late reply. I haven't had
time to chase this further.
And while I think that you may be right in that the server is abusing
the IMAP spec in some
manner, it might also be that VM isn't handling newer features
properly. The old saying of
"be strict in what you send, tolerant of what you accept" is key
here.

Fetchmail 6.3.6 has no problems pulling email off Lotus Notes using
IMAP, so I would hope that VM
would be able to do so too.

Do you have any suggestions for what I can do to make VM more
tolerant? Or can I tweak the settings to see if
there is something that will help make this reliable?

I'm just lousy in elisp hacking, not skill at all, but I'm happy to
test patches and see what happens.

Thanks,
John

Uday S Reddy

unread,
Nov 18, 2009, 1:39:02 PM11/18/09
to
John Stoffel wrote:

> Thanks for the reply, and sorry for the late reply. I haven't had
> time to chase this further.
> And while I think that you may be right in that the server is abusing
> the IMAP spec in some
> manner, it might also be that VM isn't handling newer features
> properly. The old saying of
> "be strict in what you send, tolerant of what you accept" is key
> here.

Agreed. I should have looked up what the character 10 was. It is Line
Feed!

So, you seem to have gotten a message that has LF instead of CRLF (which
is required by the IMAP spec, God knows why). I think this is probably
needless hair-splitting. I am adding a variable called
vm-imap-tolerant-of-bad-imap, which is 0 by default. If you set it to
1, it will let all these violations pass through. So, you will be able
to at least download the messages. You will still need to be watchful.
Who knows what else might break if the messages don't have CRLF's?

Please look for revision 594 (in the 8.1 branch).

I also recommend that you use the IMAP server folders instead of
traditional downloading. (See the info section on Server Folders.)
This will preserve the server copy of your email in tact even if the VM
cache might get damaged.

Cheers,
Uday

Kurt Hackenberg

unread,
Nov 18, 2009, 7:54:52 PM11/18/09
to

Uday S Reddy <uDOTsD...@cs.bham.ac.uk> wrote:

>So, you seem to have gotten a message that has LF instead of CRLF...

Maybe. I'd want to see the network traffic, and debug this for real.

>...(which is required by the IMAP spec, God knows why).

Because mail is ASCII, and that's how ASCII works. The Unix thing of
storing only LF in text files is a convenience hack, and not
compatible with other systems.

>I think this is probably needless hair-splitting.

No doubt the RFC requires sending CRLF; does it saying anything about
relaxed acceptance of variants?

>I am adding a variable called vm-imap-tolerant-of-bad-imap, which is
>0 by default. If you set it to 1, it will let all these violations
>pass through.

Lisp true/false values are conventionally t and nil respectively.
Sounds like you're thinking in C.

Julian Bradfield

unread,
Nov 19, 2009, 4:48:15 AM11/19/09
to
On 2009-11-19, Kurt Hackenberg <k...@pnnnnx.kom> wrote:
> Uday S Reddy <uDOTsD...@cs.bham.ac.uk> wrote:
>>So, you seem to have gotten a message that has LF instead of CRLF...
> Maybe. I'd want to see the network traffic, and debug this for real.
>>...(which is required by the IMAP spec, God knows why).
> Because mail is ASCII, and that's how ASCII works. The Unix thing of
> storing only LF in text files is a convenience hack, and not
> compatible with other systems.

No, that's not how ASCII works. ASCII didn't define a line end
convention (go look at the standard - it's online).
The CRLF convention arose from teleprinters, and some operating
systems adopted it, and others didn't. It's a historical accident that
the designers of TCP-based protocols decided to standardize on
CRLF. They could have made another choice.

Uday S Reddy

unread,
Nov 19, 2009, 10:15:25 AM11/19/09
to
Kurt Hackenberg wrote:

>
> Lisp true/false values are conventionally t and nil respectively.
> Sounds like you're thinking in C.

No, I am just being open to the possibility that multiple levels of
tolerance may be needed in future.

Cheers,
Uday

PS Do we really need a yet another war on CRLF's? :-)

blueman

unread,
Dec 30, 2009, 7:58:12 PM12/30/09
to
Uday S Reddy <uDOTsD...@cs.bham.ac.uk> writes:

> Kurt Hackenberg wrote:
>
>>
>> Lisp true/false values are conventionally t and nil respectively.
>> Sounds like you're thinking in C.
>
> No, I am just being open to the possibility that multiple levels of
> tolerance may be needed in future.
>

Even with multiple level of tolerances, you still would do off=nil in
Lisp...

0 new messages