meeting minutes
---------------

GRAAP-WG meeting 2 (Technical Session 1)


focus of session: negotiation aspect of WS-Agreement
----------------
Explain the term states and protocol.

We started with a repeat of some of the overview presented in the
previous session.  We discussed the roles of the parties in WS-A
(init/provider, client/service), pointed out the assymetry of the
model, and talked about negotiation operations.

Negotiation state is captured in the state of the terms which have the
following possible values:

-status: offered (initiator offers a value)/observed/ignored
-criticality: optional or required
-negotiability: value can change or not

We had much discussion about what "offered" means.  Presently it
really means any state for which agreement has not been reached.  Much
of the confusion may be about terminology, but also stems from the
assymetry of the model (who can propose a value).

We also discussed how initial values could/should come from templates
published by providers.

It was also pointed out that, at this point, we're modelling the state
of the terms and haven't truly said much about what messages put the
terms into these states.

We discussed briefly the notion that criticality may be applicable to
term groupings in addition to individual terms.

The question about initiating rengotiation was raised, and we noted
that the present model only permits an initiator to cause renegotiation.

The state space for the terms was discussed even those these states do
not map directly to what's in the current version of the spec. which
only uses two dimensions.  The full mapping from the spec. to the
logical may not be complete/correct at this point.

We again discussed the confusion of the term "offered" and resolved to
discuss it further on the list.  Other words like "proposed" or
"requested" were thrown about.

Following the state space, we worked through a state transition
diagram.  There are some ambiguities, particularly with respect to
getting into a really final state, observed-fixed from other states,
and what it means to bypass some intermediate states.

There was also concern that there was no time aspect to the overall
picture.  That is, no transitions were triggered by time.  Part of
this is that lifetime is handled for a full agreement, and this is
just for one term.

It may be helpful to restrict paths through the state diagram in some
cases to model particular negotiation styles.

Problems with denial of service were also raised.

We ran through some example cases.  In this discussion, it became
clear that we really, truly do need a commitment phase.  This is not
in the spec. now, but needs to get in there.  We take it as an area to
be worked out.

We also discussed the possibility of completely removing the idea of a
negotiability dimension when/if we start adding constraints.  The
presence or absence of a constraint could indicate the negotiability.
A separate set of constraints could be maintained for both the
provider and initiator.  They act as hints to the other party.

Renegotiation was discussed, and it is possible that we may need two
different methods, one for updating existing terms, another for
providing entirely new terms.