Graham,
GK> is there scope
GK> here for collapsing burgeoning historic thread information into a hash that
GK> can be used to confirm the sending agent's knowledge of the threads in
GK> progress.
although the typical case is for threads to be short, it is not that
uncomon to have threads that are extremely long. I think we should
attend to that occurrence.
Whether we do it through a special label (eg, hash) or through a
syntactic way of eliding some of the thread references, I don't know.
GK> Maybe that's overkill: if an agent supplies just recent generation thread
GK> history in any message it sends, I think the complete thread structure can
GK> be assembled from the various messages.
Perhaps it would be sufficient to have the thread syntax permit saying
something like a sequence:
<first message in thread>
<notation saying messages are being skipped>
<context for tail of thread>
<message being replied to>?
So, the first two items would be the equivalent of the hash, but without
extra computation.
d/
--
Dave Crocker <mailto:dcrocker(_at_)brandenburg(_dot_)com>
Brandenburg InternetWorking <http://www.brandenburg.com>
Sunnyvale, CA USA <tel:+1.408.246.8253>, <fax:+1.866.358.5301>