Closed cabo closed 2 years ago
How about:
representation: a representation of a data tree [RFC7950]. YANG currently has three representation formats: based on XML [RFC7950], JSON [RFC7951], and CBOR [RFCthis].
representation node: a schema node or an instance of a schema node that is part of a representation. This includes all data node instances, as well as most schema nodes, with the notable exceptions of "choice" and "case".
IMO it is a big mistake to omit choice and case schema nodes from numbering
1) very confusing to invent this artificial subset of schema node 2) it is very useful when parsing or encoding to start with the parent of the expected nodes 3) It is useful to do a GET on a choice or case, rather than the N possible GETs on the accessible descendants 4) it causes code to handle these 2 schema nodes in a special way (e.g. assign SIDs anyway)
YANG defines "data tree" which matches
o data tree: An instantiated tree of any data modeled with YANG,
e.g., configuration data, state data, combined configuration and
state data, RPC or action input, RPC or action output, or
notification.
"If we include case and choice, then we can use the term schema node."
@abierman to create example?. No, just need to fix ietf-system example.
One use case for CoMI is the ability to generate client/server code that does not need to store name strings, only SIDs. In this type of implementation, all the schema item SIDs will be needed, even though choice and case will not be on the wire.
Closed by #136
We probably could use a word for schema nodes that we encode, i.e.,
(data node +) action, rpc, input, output, notification
Schema node also includes choice and case, which we aren't interested it.