Hi Matt,
This is interesting. I have a few observations (looking from the perspective of our needs for deriving a VDU node for NFV profile):
1. tosca.capabilities.compute derives from tosca.capabilities.container - but tosca.capabilities.container is empty. What is the point - do you expect to add properties later?
2. tosca.nodes.container node does not exist. There is a tosca.nodes.container.Runtime, a tosca.nodes.container.Application. Is this on purpose, or is something more to come?
3. tosca.nodes.compute derives from tosca.nodes.Abstract.Compute, not from tosca.nodes.container (correct, since tosca.nodes.container is not defined, as per above).
4. tosca.nodes.Abstract.Compute does not have the tosca.capabilities.container, but the tosca.capabilities.compute.
In essence, the changes may address a set of issues, but not the issues we have in NFV profile with the VDU.Compute node (which at present is derived from tosca.nodes.compute, with a lot of deprecations and overlaps of capabilities, because what we need to add). Even if we were to derive from tosca.nodes.Abstract.Compute, while there are less deprecations, we still have the issue with inheriting the tosca.capabilities.compute which creates duplication with the capabilities we need to add.
I would have been inclined to derive VDU from tosca.nodes.container, if such node would have been defined.
Best regards,
Michael