Broaching the idea of ANNOTATE-LESS (a.k.a COMMENT)

[Dan Karp <dkarp@xxxxxxxxxx>]

I've read through the latest ANNOTATE draft, and unfortunately it appears that there's no way for us to implement it without some *really* major rearchitecture.  Some of the things that are really non-trivial for us to do include:

   - SORT on annotations
   - SEARCH on individual annotations
   - (possibly) public/private annotations

I thought about it for a while, and I realized that the much simpler set of things we *can* implement still makes a pretty compelling subset of ANNOTATE functionality.  So I was wondering if it might be OK to try a second draft (called COMMENT or somesuch) encapsulating the following functionality:

   - one COMMENT annotation field per message
   - the COMMENT annotation maps directly to the "/comment value.shared" ANNOTATION (and hence tags along via COPY)
   - a COMMENT annotation must be valid UTF-8 text or NIL

   - the COMMENT annotation is SEARCHable, settable via STORE and APPEND, and retrievable via FETCH
   - the COMMENT annotation is non-SORTable
   - unsolicited notifications include the full COMMENT value

   - COMMENT must be explicitly enabled via SELECT (COMMENT), EXAMINE (COMMENT), or ENABLE COMMENT

So the syntax would look something like this:

   C: A001 FETCH 1 (UID COMMENT)
   S: * 1 FETCH (UID 7723 COMMENT NIL)
   S: A001 OK FETCH completed

   C: A002 STORE 1 COMMENT "This is the message's comment!"
   S: * 1 FETCH (COMMENT "This is the message's comment!")
   S: A002 OK STORE completed

   C: A003 FETCH 1 (UID COMMENT)
   S: * 1 FETCH (UID 7723 COMMENT "This is the message's comment!")
   S: A003 OK FETCH completed

   C: A004 STORE 1 COMMENT.SILENT "Updating the comment..."
   S: A004 OK STORE completed

   C: A005 APPEND Sent COMMENT "Attached to new message" {14432+}
   C: ...
   S: A005 OK APPEND completed

   C: A006 SEARCH COMMENT "comment"
   S: * SEARCH 12 44 156
   S: A006 OK SEARCH completed

I spoke with Cyrus about this, as he's one of the authors of the ANNOTATE extension and hence this is fundamentally his (and Randall's) work.  He suggested taking it to the full IMAPEXT list, and I've obliged.