On Wed, 10 Dec 2003, Arnt Gulbrandsen wrote:
> I saw Mark's comment about RFC 1176, and disagree. That's IMAP version
> 2, I wouldn't worry about it.
Wrong!!
The IMAP protocol is carefully designed to provide not to break clients
when servers are upgraded, and to negotiate newer facilities with clients.
Port 143 is assigned to IMAP2. IMAP4 and IMAP4rev1 are permitted to use
port 143 only because they maintain compatibility with IMAP2 clients.
Any server which sends any new protocol responses without the client
requesting such is broken.
The only legitimate reason for an IMAP4rev1 server not to work with an
IMAP2 client is if the server disallows the LOGIN command.
> Either you should support IMAP2, in which
> case you should actually test with IMAP2 clients (good luck finding
> any), or you should decide not to support it and your current responses
> are fine.
That is an amazingly callous thing to say.
Many people in the world still use Pine version 3.xx, which is IMAP2bis
(IMAP2 + FETCH BODY + CREATE/DELETE/RENAME/APPEND). Among other things,
it's one of the few clients that runs on DOS and 16-bit Windows (don't ask
me why people still want to use such things -- I don't know!).
We have too many broken servers already. There is no reason to inflict
deliberate brokenness on the world. There is no reason to go to the extra
work of sending a UID when the client did not ask for it.
A UID is sent only in the following cases:
UID FETCH
UID STORE
FETCH with UID as one of the items to be fetched
A UID is NOT sent in the following cases:
FETCH with UID not as one of the items to be FETCHED
STORE
unsolicited FETCH responses
-- Mark --
http://staff.washington.edu/mrc
Science does not emerge from voting, party politics, or public debate.
Si vis pacem, para bellum.
