Move Collection Section out of the Core Spec

Authors

Luc Moreau, 19 June 2009.

Subject

OPM 1.01.

Background

At the first OPM workshop, it was recognized that collections are important. The original spec OPM1.00 didn't say anything about collections. It was revised by introducing section 9 in OPM1.01.

Problem addressed

_Section 9 was only intended as an illustration of what can be done in OPM to deal with collections. It was in no way prescriptive and complete. Since then, it has become clear that the concept of specialization, i.e. OpmProfile, is becoming important in OPM._

Proposed solution

_Remove section 9 from the OPM specification. In the specification, introduce the concept of OpmProfile, and refer to emerging effort to create an OPM profile for collections._

Rationale for the solution

Keep the core specification focused on core issues, and use profiles for OPM specialisations. Also, allow a faster development cycles of profiles if appropriate.

Comments

Community is invited to provide comments on proposals.

comment 1 ben clifford

I agree with this proposal.

comment 2 by Simon Miles

I also agree that this seems sensible.

-- JimMyers - 17 Sep 2009

I'm not sure I see what the current section 9 says beyond that accounts of collections running through processes can be refined to show the processes that separate the components and reassemble them. There is one bit that starts to formalize the notion of collection by connecting the idea of roles on the aggregation/disaggregation operations to accessor methods on the collection which is 'cool' but without more work I'm not sure of its practical use - does anyone want to crawl the subgraph of OPM to get the collection elements? Why not get them directly as the output artifacts of the diggregator process?

So - I'm in favor of dropping this until we have a more complete notion of collections and the rules that might govern how one can infer relations in OPM (what does the provenance of a collection imply about the provenance of its elements, what can one infer about the collection from the provenance of its elements?).

comment from Joe Futrelle

Modularity is good.

Vote

Luc Moreau, yes

Paolo Missier, yes

Jim Myers, yes

Simon Miles, yes

Yogesh Simmhan, yes

Joe Futrelle, yes

Paul Groth, yes

NataliaKwasnikowska, yes

Jan Van den Bussche, yes

Eric Stephan, yes

Outcome

Unanimously adopted (Yes: 10/ No: 0)
to top