Should the Guidance doc recognise existing, non-interoperable, profiles as valid

[nicholascar]

No description provided.

[kcoyle]

Notes from the minutes of the meeting:

there are existing things that are called profiles in the wild. Does the guidance document recognize those
20:42:36 [roba]
Should guidance about profiles identify existing practices using implicit profiles as valid examples or as prior work only
20:42:38 [kcoyle]
... as valid profiles, or non-conformant prior work?

(Aside: non-interoperable or non-conformant with what?)

[dr-shorthair]

Of course they are valid! However, existing profile documents might not be formulated so rigorously...

A key thing here is to be clear about the distinction between an instance of a profile as the conceptual thing, and one or more cencrete artefacts which express and support it, including the guidance documents, schemas ontologies and shapes documents, controlled vocabularies, etc. Many legacy profiles will be expressed in a single artefact. Those are still valid profiles.

[aisaac]

I'm not sure I understand this issue. Trying to rephrase: do we mean that there are profiles that exist prior to our recommendations and that do not follow them, and the questions are whether (1) we agree to name them profile and (2) we seek to judge them as valid or not?

On (1) I'm keen to recognize as profiles everything that matches our definitions even a bit losely - especially if it names itself a profile.

On (2) I think it's too early to judge. I would wait until we can see how strict (in terms of MUST and SHOULD) our recommendations are.
This said, I can already anticipate that for example we may want to say that some profiles are not very good web citizen because they don't have a URI.

[rob-metalinkage]

Its not the profiles which are non-interoperable - its the descriptions which cannot be interoperable without a canonical description formalism..

so we can retrofit an interoperable description to any prior profile work - so long as those profiles conform to our semantics.

A "lump of green putty" is not a profile.. a "specification inspired by but not actually interoperable with another spec" is not a profile (even if it wants to use the word in title or description).

Its not our problem to recognise pre-existing profiles - but we can describe those that behave according to the accepted definition.

[aisaac]

@rob-metalinkage I think I see where you're going to: you would like to describe 'old' profiles with our apparatus and see whether that works? If yes, then this sounds good to me. I guess this is what I had in mind for some of the 'example/implementation' stuff at the end of the Profile Guidance doc.

[rob-metalinkage]

@aisaac that is my Use Case for implementation - I need to describe all the existing OGC specifciations, including profile relationships, and then set up tooling for communities to publish more specific profiles as they need them... just waiting for ontology to hit some visible status - maybe 2PWD - or even FPWD if we are not getting requests to change - so I can implement and get feedback from that community.

[aisaac]

@rob-metalinkage this sounds very good, and I'd be happy to include (at least a part of) it in the Guidance doc.