RE: Is there are replacement for 'UIDL'?

> -----Original Message-----
> From: Paul Smith [mailto:[email protected]] 
> Sent: Tuesday, February 19, 2002 2:31 PM

> Basically, if you send their server a 'UIDL' command it gives 
> you a set of IDs for the messages, then if you ask a few hours later, 
> it'll give you a totally different set of IDs for the same messages.
> As far as I am concerned this is a breach of the relevant bits of
> RFC 1939.

This is indeed bad - the UID for a message should not change.

> ... they say that their server does not add a UIDL AT 
> ALL to any messages on their servers, they are using, 
> instead, their own Unique Message ID [which does persist across
> sessions].

That's fine - there's nothing that says that the UID has to be stored
*on* the message, as long as the same UID is returned each time you ask.
If you're storing messages in a database, you might use the primary key
of the message in the DB as it's UID, for example.

> ... <they> blame BTWebworld [the domain-forwarding source of most of
my mail] 
> for ?somehow attaching changeable UIDL's which are confusing VPOP3
[our 
> mail software].

I have seen an X-UIDL header added to mail, which can lead to some real
problems - in particular, if a server accepts that as the UID of the
message, you can send it multiple different messages with the same
X-UIDL header and end up with multiple messages with the same UID on the
server.
 
> Is there anything I've missed, or is it just waffle from an ISP which 
> doesn't know what it's talking about?

It does sound like a waffle.  If they assign an invarient unique ID to
each message and return it as the UID of that message, then the UID of a
message should never change.  If they admit to the UID changing, then
they either (1) don't do what they say they do or (2) have a bug.
Besides, once a message is in their system how do they explain their
claim that the forwarder is somehow responsible for the changed UID?
*They* are now in control of the message!
 
-- jeff