What are pros and cons to a 'core set' of Parameter Constraints?

[garethsb]

The draft spec currently lists a few of the most important Parameter Constraints that the activity group has identified, for video and audio streams:

• urn:x-nmos:cap:format:media_type
• urn:x-nmos:cap:format:grain_rate
• urn:x-nmos:cap:format:frame_width
• urn:x-nmos:cap:format:frame_height
• urn:x-nmos:cap:format:channel_count
• urn:x-nmos:cap:format:sample_rate

Including a 'core set' in the specification has the advantage that Controllers have to do something to claim support with the spec. It means that a future test suite for Controllers can report a WARNING if they ignore these basic constraints.

Another upside is that these constraints could be included explicitly in the spec schemas, allowing Receiver implementations to be more easily checked to be valid by Controllers and the Testing Tool.

One downside is that there would therefore be some duplication between the spec/schemas and the proposed new Capabilities register in the NMOS Parameter Registers.

Another downside is that it could enshrine multiple tiers of compliance, perhaps making it less likely Controllers will implement support for other constraints the activity group has identified, or that are added to the register in the future?

Thoughts?

[garethsb]

Core urn:x-nmos:cap:meta: attributes could also be added to the spec/schemas, for example label.

[garethsb]

Consensus during the 2-Nov-2020 meeting was that:

• on the one hand, it made sense to have the meta items (label and preference) defined in the spec and schemas of BCP-xxx-01 v1.0
• however, it would be better if the parameter constraints themselves were only defined by the Capabilities register in the NMOS Parameter Registers (see https://github.com/AMWA-TV/nmos-parameter-registers/pull/21) as the issue of current/future "useful sets" and adhoc items, versioning etc. is a real concern

Therefore #6 should be closed and instead the existing 'core set' should be removed from the spec itself. And #9 should be rebased and merged.