next rev model

[Greg Pavlik <greg.pavlik@oracle.com>]

Gentlemen -- another draft that attempts to answer Peter's questions as 
well as reflect what I think was the consensus outcome of the 
discussions we had.  It's clear that some folks are going to challenge 
us sharply on the utility of "pass by value" (what I consider to be the 
normal model for simple transaction protocols) as well as the need for 
exposing nesting relationships in the context structure; actually 
Peter's latest email suggests that the whole idea of context should be 
discarded altogether in favor of SOAP headers. So....

Here's a short list of TODOs. Please feel free to create a better list 
or, should we stick with this, swap TODOs. Can we

1) make sure that this is an accurate synopsis of our shared 
understanding? (Mark/Eric)
2) buttress it with some examples (Eric)
3) take a stab at building up the defenses from the likely objections 
(why nesting structures are useful to expose in the context; what 
contexts have in common  at the most basic level that needn't be 
reinvented continuously in every protocol that uses SOAP headers, why 
pass by value is useful, etc.) (Mark)

It would be great if we could have feedback from everyone by tomorrow 
afternoon.

Thanks.

Greg

CAF model clarification v3.doc