nmh-workers
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Nmh-workers] IMAP/nmh, again


From: Paul Vixie
Subject: Re: [Nmh-workers] IMAP/nmh, again
Date: Fri, 27 Oct 2017 00:41:58 +0200
User-agent: Postbox 5.0.20 (Windows/20171012)



Bakul Shah wrote:
On Thu, 26 Oct 2017 17:26:37 -0400 address@hidden wrote:
address@hidden writes:
AIUI, the big issue has always been that nmh has expected message numbers
to remain static until explicitly changed (i.e.  message 35 *stays* message 35
until 'folder -pack' or something changes it), while IMAP message numbers
can change even during a connection, so UUID's need to be used instead,
which means keeping a message<->uuid mapping someplace.

Indeed.  Keeping such mapping is what I planned to do. folder
-pack only changes the local msgID<->UID map so no need to
talk to the server for that. But as UIDs are assigned by the
imap server, after a refile there will be messages with no UID
(at least temporarily).

fwiw, that sounds pretty reasonable.

And making that robust under concurrent access sounds even worse...

Luckily this is taken care of by various imap extensions as
the problem exists even for "pure" imap clients. But I have a
feeling we will discover more corner cases + imap server
implementations that don't quite do the right thing.

there's a think in imap called "push", which is part of why i keep saying that true imap support in mh would look like prayermail's session layer, although probably implemented as something like gpg-agent or ssh-agent. we're going to need a persistent connection, regardless of how fast or slow it is to set up and/or tear down those connections, be then LAN, WAN, SSL, TLS, or whatever.

if ssh can talk to ssh-agent over a unix domain socket, and gpg likewise to its gpg-agent, then the unix way of doing this is clear. it's possible that a simple persistent imap session, accessible via reconnections to the unix socket, is all we need. or, it's possible that something more like prayermail, where there's some session layer caching and a different non-imap protocol spoken between the agent and the various mh commands, would serve us better.

sadly, i can already think of non-imap-related reasons why mh needs an agent of this kind. i think i'm infected with non-unix thinking. ouch.

--
P Vixie




reply via email to

[Prev in Thread] Current Thread [Next in Thread]