See also: IRC log
Present: alewis, dbooth, Marsh
Regrets:
Chair: dbooth
Scribe: alewis
Scribe: Amy argues in favor of removing notations for faults
... Jonathan suggests that explicit characterization may be a useful exploration of whether the suggested rulesets are appropriate/complete
Scribe: Examining patterns 2, 2a, and various other specializations of 2, it is noted that the model notation language implies stronger separation between output destination/output fault destination than the language supplied.
<Marsh> Amy: Might be a problem. Some of our patterns distinguish when some nodes do NOT overlap.
Scribe: Jonathan asks if a node is a node (that is, always a separate thing). David suggests that each node in the graphic representation represents a role, but may
... David suggests that each node-indicator in the graphic representation represents a role, but may represent the same node (or set of nodes).
... David suggests that where it is necessary to distinguish between *nodes* rather than between *roles*, we may need to use the prose.
<Marsh> Amy: If we define the circles as roles, instead of nodes, we start to run into the problem of distinguishing between classic request-response vs. bounce request-response, where one node...
... ... submits request, another recieves response.
... ... David calls bounce request-response third-party request response.
... ... look at P2e.
... ... The circle indicates node identity.
... ... That's what distinguishes this from P2d.
... ... We'll have to refine the definition of what the circles represent, roles or nodes.
... ... If we need to establish that a role overlaps with another node, we need to indicate that.
... ... Dotted circle for a role?
... David: You use a different notion of "role".
... ... I use it synonymous with "variable".
... ... In P2d, A, S, and B are all variables.
... Amy: P2c, d, e have the same thing going on. Why one has 4 variables, 3, or 2.
... David: Less constrained. P2d is the same as P2c with A=C
... ... P2d: There exists some software A, some software S.
... ... not constrained to be separate pieces.
... ... There is a message sent from A to S.
... There is a subsequent fault message sent from S to A, or an output message from S to B.
... ... Haven't said software A must be different than software B.
<dbooth> ACTION: dbooth to clarify in Notation that if nodes are different that does NOT mean that the nodes are REQUIRED to be different
<Marsh> Amy: Immediate visual impact communicated is different than what your intent is.
... ... Possibly resolved by giving nodes not described as distinguished fuzzy edges.
<dbooth> ACTION: dbooth to add a p2d1 that returns the fault to party B
<Marsh> [Some discussion about notation, whether it matches the prose.]
... David: Are we comfortable with the notation? A blue circle representing a role?
... ... Don't see how to fix the notation. We need the concept that parties may be different. Can't imagine we need to express that parties MUST be different.
... Amy: OK.
<dbooth> http://dev.w3.org/cvsweb/~checkout~/2002/ws/arch/glossary/wsa-glossary.html#mep
<Marsh> Amy: Do we need to sync up the definition of MEP between part 2 and the arch glossary?