Proposal: rationalization of faults in patterns

I'm sending this to the main list, rather than the MEPs task force,
because I believe the issue raised is orthogonal to those of concern to
the task force.

Problem: the current set of patterns (part two) is more complicated than
necessary, because it models individual faults in the message flow.

Proposal: instead of modeling each fault, create two standard rule sets
for fault generation.  Each pattern then MUST express its adherence to
one rule set or the other (note: more than these two may be useful, but
I *believe* that these two cover all the necessary ground).

Ruleset 1: a fault message MAY replace any message after the first in a
pattern.  (Note: all patterns except 7 currently use this rule).

Ruleset 2: any message in a pattern MAY trigger a fault message in
response.  (Note: pattern 7 uses this rule).

If approved, I would like to see an action that the editors make changes
corresponding.  I'd like to invite discussion, especially if I haven't
been clear.

Amy!
-- 
Amelia A. Lewis
Architect, TIBCO/Extensibility, Inc.
alewis@tibco.com

Received on Wednesday, 28 May 2003 14:22:32 UTC