On Thu, 2009-12-17 at 11:29 -0800, Mark Crispin wrote:
> On Thu, 17 Dec 2009, Timo Sirainen wrote:
> > On Thu, 2009-12-17 at 10:22 -0800, Mark Crispin wrote:
> >> No server "caches the flags" as you describe.  If the mail store in the
> >> server supports multiple simultaneous sessions to the same mailbox (e.g.,
> >> mix and mbx formats in UW), then flags are updated immediately.
> > Well, not necessarily immediately, at least not all servers. A NOOP
> > command should refresh flags and other things with most servers.
> 
> A server that requires use of the NOOP command to synchronize is broken.

Sure, but there are such servers. For example Cyrus. Actually in Cyrus
it seems to work in a rather weird way:

session 1: 1 store * +flags \seen
(session 2 won't see this change, no matter what it does.)

session 1: 2 noop
(session 2 still won't see the change with fetch flags.)

session 2: 3 noop
(now session 2 finally gets the untagged FETCH FLAGS reply.)

> A client that uses the IDLE command has no reason to use the NOOP command, 
> ever.

Yes, IDLE and NOOP are the commonly working ways. With the above Cyrus
example session 2 could have been IDLEing instead of doing a NOOP.

> > Hmm. Dovecot's current behavior is that doing a FETCH FLAGS just after
> > its flags had changed in another session first returns one FETCH FLAGS
> > with the old flags (the FETCH reply), then another FETCH FLAGS with the
> > updated flags (after it noticed the changes).
> 
> That's OK.  It's in the same set of responses, isn't it?  The later flags 
> response overrides the former one.

Right. I changed it anyway, maybe it'll help some poorly implemented
client.

Attachment: signature.asc
Description: This is a digitally signed message part

_______________________________________________
Imap-uw mailing list
[email protected]
http://mailman2.u.washington.edu/mailman/listinfo/imap-uw

Reply via email to