Re: [Netconf] IETF Last Call Gen-ART review of draft-ietf-netconf-restconf-15

[[e-mail] (Dale R. Worley)]

Andy Bierman <[e-mail]> writes:
>>     8.  RESTCONF module
>>
>>              The following data-def-stmt sub-statements have special
>>              meaning when used within a yang-data-resource extension
>>              statement.
>>              - The list-stmt is not required to have a key-stmt defined.
>>              - The if-feature-stmt is ignored if present.
>>              - The config-stmt is ignored if present.
>>              - The available identity values for any 'identityref'
>>                leaf or leaf-list nodes is limited to the module
>>                containing this extension statement, and the modules
>>                imported into that module.
>>
>>     It seems like poor practice to have the extension be described as
>>     changing the semantics of Yang.  Better would be to turn these into
>>     constraints, so that the valid contents of yang-data are a subset of
>>     Yang, but that subset has the same semantics as Yang prescribes:
>>
>>              - The if-feature-stmt must not be present.
>>              - If the config-stmt is present, its value must be 'false'.
>>              - The available identity values for any 'identityref'
>>                leaf or leaf-list nodes is limited to the module
>>                containing this extension statement, and the modules
>>                imported into that module. [unchanged!]
>>
>>     The item "The list-stmt is not required to have a key-stmt defined."
>>     is redundant, since everything inside yang-data is not configuration
>>     data, and non-configuration lists need not have keys.
>
> The suggested edits are not accepted.
> They prevent groupings from being used within a YANG data template.

That bit about groupings is a subtle point I hadn't thought of.  If the
WG is OK with that, I'm OK with that.

Though I think the item "The list-stmt is not required to have a
key-stmt defined." is still redundant -- everything inside yang-data is
not configuration data, and non-configuration lists need not have keys.
Even if the list is inside a grouping, that doesn't change anything, in
that the list has a key-stmt or not depending on how the grouping is
defined, and either case is valid (in regard to the data template)
without a special rule.

Dale