With respect to the Mount draft. I like the goal of being able to take an assortment of device models (D1..Dn) and merge them in a scalable manner into a single aggregate model (A).
My use case is a management interface aggregator. A sort of middle box with multiple netconf interfaces on the managed device side and one netconf interface on the manager side. In my case, one that performs persistence and pre-provisioning in addition to aggregation. The draft proposes to do this by adding a new syntax 'mount' along with supporting semantics. While adding yet another syntax to Yang is a possibility, I'd like to propose something else. Specifically, I'd like to adjust the semantics for 'uses', 'rpc' and 'notification' and keep the appearance of the current syntax. 1) Allow the 'uses' verb to use a module in addition to groupings. 2) Allow the 'rpc' and 'notification' nouns to be used in other places in the schema tree besides at the top module level. Given these features, I see them being used three ways. First to mount D on A by modifying D, write augment A. Second to mount D on A by modifying A, write uses D. Third to mount D on A with a third separate module, write augment A uses D. At a protocol level, I guess data would just be deeper in the tree. Rpc and notification would now need a prefix to their name to show where in the tree they are from. I wonder if others will see this as a more consistent and general purpose adjustment to Yang/Netconf.
_______________________________________________ netmod mailing list [email protected] https://www.ietf.org/mailman/listinfo/netmod
