Re: UIDL Question

> sheryl.bruggy-CK+cXun/dNAQrrorzV6ljw@xxxxxxxxxxxxxxxx wrote:
> > Hi All,
> >
> > I have received a complaint from a vendor who supplies our organisation with 
> a CRM product.  It seems that when the CRM application downloads messages via POP, 
> it sees the UIDL for these all these messages as being the same.

> They don't look "all the same" to me.  Perhaps the vendor has a bug in how they
> parse the value?

> > I did a test by telneting to our server and saw the following results:
> >
> > UIDL
> > +OK unique-id listing follows
> > 1 2579-1189132072
> > 2 2580-1189132072
> > 3 2581-1189132072
> > 4 2582-1189132072
> > 5 2583-1189132072
> > 6 2584-1189132072
> > 7 2585-1189132072
> > 8 2586-1189132072
> > 9 2587-1189132072
> > 10 2588-1189132072
> >
> >
> > A quick search of the docs only revealed the following:
> > "The UIDL implementation is based on IMAP4 Universal Identifiers."
> >
> > Does anyone know how the IDs are generated?

> It looks like the IMAP UID of the message (with also happens to be the .msg file
> number) and the IMAP UID VALIDITY of the folder (which happens to be the unix
> time when the folder was created).

That's what our POP3 UIDs look like. More generally, POP3 UIDs are allowed to
be almost any string as long as the string is unique across the server. This is
quite unlike IMAP4 UIDs, which are integers in a strictly ascending sequence
from first message to last.

Any POP3 client has to deal with the full generality that the specifications
allow for these things or it is incompliant. And since there are
implementations out there that use hash values and not nubers, ours is hardly
one of the more difficult ones to deal with.

                                Ned