Re: Exclusive-access locking of maildrop

Michael D'Errico <Mike@software.com> Wed, 01 June 1994 21:17 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa10574; 1 Jun 94 17:17 EDT
Received: from CNRI.RESTON.VA.US by IETF.CNRI.Reston.VA.US id aa10570; 1 Jun 94 17:17 EDT
Received: from PO3.ANDREW.CMU.EDU by CNRI.Reston.VA.US id aa00264; 1 Jun 94 17:17 EDT
Received: (from postman@localhost) by po3.andrew.cmu.edu (8.6.7/8.6.6) id RAA02776; Wed, 1 Jun 1994 17:12:25 -0400
Received: via switchmail for ietf-pop3+@andrew.cmu.edu; Wed, 1 Jun 1994 17:12:24 -0400 (EDT)
Received: from po3.andrew.cmu.edu via qmail ID </afs/andrew.cmu.edu/service/mailqs/q002/QF.UhvDZY:00UdbA:WE57>; Wed, 1 Jun 1994 17:11:03 -0400 (EDT)
Received: from rome.software.com (rome.software.com [198.17.234.2]) by po3.andrew.cmu.edu (8.6.7/8.6.6) with ESMTP id RAA02686 for <ietf-pop3@andrew.cmu.edu>; Wed, 1 Jun 1994 17:10:47 -0400
Received: from rome (rome.software.com [127.0.0.1]) by rome.software.com with ESMTP id AAA79 for <ietf-pop3@andrew.cmu.edu>; Wed, 1 Jun 1994 14:10:40 -0700
To: ietf-pop3@andrew.cmu.edu
Subject: Re: Exclusive-access locking of maildrop
Date: Wed, 01 Jun 1994 14:10:39 -0700
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: Michael D'Errico <Mike@software.com>
Message-ID: <19940601221040.AAA79@rome.software.com>

I originally wrote:

mike> I suggest changing to spec. to say that a server MAY
mike> acquire an exclusive-access lock if it is desirable, but
mike> that it is not required.

Steve Dorner replied:

sdorner> I'm terribly worried about removing the locking requirement.
sdorner> If you do not require that the server 'lock' the 'maildrop',
sdorner> then you will have implementors who do an IMAP sort of thing,
sdorner> where messages come and go while the client is examining the
sdorner> maildrop.

I see your point, and I agree that this is undesirable.  But what do
you do if you have a maildrop that can be accessed by either a POP
server or an IMAP server?  Do you prevent the IMAP server from opening
the maildrop if the POP server has it open?  Or do you just disallow it
from changing the contents?

Either way, the IMAP server has to be "POP-aware" which means there
needs to be a standard way to lock maildrops, etc. so that different
implementations can interact correctly.  This would require changing
both the POP and IMAP specs. and I doubt we'd ever come up with a
universal way to lock maildrops.  The only other ways to enforce this are
to force people to use POP and IMAP servers from the same implementor
(assuming that the servers cooperate) or to pick one protocol over the
other.

Alternately, we can allow multiple POP and IMAP servers to operate on
the same maildrop, along with all the problems that this would cause.
It would be better to have it documented in the POP spec. that "messages
may disappear due to simultaneous POP or IMAP sessions" than to have
them disappear with no advance warning.

I don't feel too strongly about this, but I see a conflict that could
cause problems, so I would like to see this change made.

Michael D'Errico
mike@software.com