Luc Moreau, June 15, 2009
OPM v1.01
Writing a formal specification for OPM is a time consuming activity, since changing a "constraint" or "definition" in OPM may require the whole formal specification to be changed. For instance, at the PC3 workshop, we discussed the possibility of removing a constraint about the number of wasGeneratedBy edges for a given artifact in a given account. A change in that constraints may result in many proofs and properties to be revisited.
It is a near impossible task to maintain a specification of OPM and a formal specification in sync under a tight schedule.
Make the formal specification a separate document. Use plain English and pictures in the main specification to define OPM.
This will give us a chance of a standalone definition of OPM without theoretical inconsistencies, by the end of the year.
Community is invited to provide comments on proposals.
I agree entirely with the sentiment and intent of the proposed change. I'm not clear exactly which parts of the specification will be removed by the change.
Essentially: remove section 5 (replace any useful content by text), express inferences in English (section 6), section 7 to be text based,
Defnitely agree, however surely this leaves us with the need to define another process to resync the two, "at some point?
And do we also specify which is the normative and which is the informative document in order to be "OPM compliant"? Usually, the formal spec would be normative and the "English" spec informative, but in our case, it might be the inverse.
Fine but realize that this means that some proposals made informally to change OPM might turn out to be ill-founded when trying for formally define them. The interaction between informal and formal version will hopefully remain.
Luc Moreau, yes
Paolo Missier, yes
Simon Miles, yes
Yogesh Simmhan, yes
Paul Groth, yes
NataliaKwasnikowska, yes
Jan Van den Bussche, yes but see my comment
Proposal is adopted (yes: 7/ no: 0).