--On Saturday, January 11, 2003 10:41 AM -0800 Mark Crispin <[EMAIL PROTECTED]> wrote:

On Sat, 11 Jan 2003 13:39:16 -0500, Cyrus Daboo wrote:
[proposed improved algorithm]
This should probably be done as a
separate THREAD=XXX option.
To make things unambiguous:

I strongly support this.  My opposition is solely to the idea of breaking
the existing THREAD=REFERENCES algorithm.
I'm not interested in breaking THREAD=REFERENCES. I'm merely saying that if the current THREAD=REFERENCES is broken, we shouldn't publish it as a standards-track RFC. Current servers don't have to drop support, but there's no reason to publish both THREAD=REFERENCES and the fixed THREAD=XXX.

I guess I agree with Mark that there's probably no need for a seperate spec that returns only threads with unseen messages. (The underlying question is whether THREAD or SORT's O(n) size responses is acceptable. I'd rather not reopen that question right now.)

Larry

Reply via email to