Re: FW: TBTF: Trace Bindings - wire view v requester/responder vi

Hi Stuart,

> Glad you find it interesting... I hope you've had a chance to glance at
> least at the second message that I send to Noah (yesterday) [1], it gets to
> more directly address the question in his response [not archived]

Yes, that helped, thanks.  But I still wouldn't claim to be able to
confidently review it for accuracy.

> Indeed process algebra expressions can be executed... you can drop
> expressions into environments that will develop traces, check for livelock,
> deadlock, safety... and also let you explore by poking them with events.

Right.  Do we have any plans to do that as a WG?  Or are we planning to
rely on human review for accuracy?

> I am very open to the tabular presentation being consider by the group as
> the more accessible - the process algebra can be developed from the table
> and for those that might want to execute it that can be done. The tables do
> contain some prose, so clearly, direct execution of the table is not really
> an option.
> My main aim in raising this as an alternative was to ask the question of
> whether it enables us to give a more compact presentation of our material
> without, in the process, making it inaccessible.

I think that it less accessible, yes.  If it could be sufficiently
reviewed or executed & verified, then what we could do is make it
normative, but also go with either the tabular or prose format in
an informative manner.  This would allow developers who didn't want
to use the formal model to implement from the informative text, but
would also provide an authoritative, testable, correct model from
which ambiguities in the informative description could be resolved.
Some might balk at that (including some developers who'd be concerned
about working from informative text), but I believe it to be a
practical compromise.

Mark Baker, Chief Science Officer, Planetfred, Inc.
Ottawa, Ontario, CANADA.

Received on Sunday, 20 January 2002 18:04:42 UTC