RE: Let's fix THE PROBLEM

Shane Dempsey wrote:
> Having visited the WISEN website...
> Through a quirk of fate IM has evolved as one of the 
> first technologies to justify the development of ISEM.
	I think the 1998 effort to get a Notification or PubSub Working
group in the IETF is probably a good example of both how things should
happen in the IETF as well as a great example of how not to do them. It
is a great example since the folk driving the effort went to a great
deal of effort to gather and present information on the scope of the
problem to be addressed prior to proposing that the WG be created. They
did much more preliminary work than I think is done for most WGs. On the
other hand, what they identified was a very BIG problem that would be
addressed by the WG and following in IETF tradition, they did not
present a plan of attack that showed how the problem could be decomposed
into bite-sized issues that could be dealt with in turn. The result,
which I've gathered from talking with a number of folk who were involved
at the time, is that people were generally scared by the size of the
problem and couldn't see how a single WG could address it.
	The reality is that PubSub is a fundamental pattern of network
communications and is implemented in some form not only in IM but also
in many other applications (database replication, printer status,
etc...) Just as we have seen that a vast number of applications were
enabled by implementing HTTP as a basic "request/response" pattern, we
would see a tremendous number of applications enabled if we were to
define and deploy a broadly usable, Internet scale, publish/suscribe
protocol.
	However, just as HTTP is not the "best" request/response
protocol that we could have designed, the PubSub that will come from the
IM efforts will also undoubtedly end up being suboptimal because it is
overly influenced by the specific requirements of just one application
domain.

		bob wyman




  [reminder: [email protected] for non-technical discussions, please]