Summing up, UID in subject

[Bernhard Reiter]

On Tuesday 13 July 2004 10:49, Bo Thorsen wrote:
> Stuart has requested that the incidence UID is put in the subject.

After the further discussion,
I also got an agreement by Martin in personal communication.
Bo and Joon already agree.
On the 13th I also agreed, there was the summary
in my email:

Date: Tue, 13 Jul 2004 18:37:44 +0200
Message-Id: <200407131837.47538.bernhard at intevation.de>

which I extend now:

|Pro aguments for the UID in the subject:
|        - possible optimisations in rare cases
|               and we don't know how often those cases occur
|       - we need something useful for the subject anyway

        - helps the current horde code to get a solution
           that works (but is slow)

|Contra arguments:
|      - it allows to implement an uncessary slow solution searching
|        the subject all the time when you could try the imap id first.
|        If clients do that, scalability of the server might go down and 
|        it might be blamed on the Kolab concept.
|      - it feels strange to have this duplicated information in the subject

Again the disadvantages seem to be outweighted by the advantages.
Especially the slow solution by design can be avoided if we inform
client developers about this danger and point to the better algorithms.
Martin proposed to have a hint (or a pointer) about this in the spec
saying that this SHOULD (as opposed to MUST) be avoided.

David: Can you also add this to the spec?

Bernhard