Re: [netconf] Create IANA-defined modules?

[Kent Watsen <[e-mail]>]

Hi Andy,

> There are a couple comments I left out of the YD review…

No problem.


> 1) if-feature usage in identities
> 
> IMO YANG conformance is still very spotty at best and conformance for identityref data nodes
> is especially broken.  I realize that if-feature is the standards-track solution to
> make a value optional-to-implement, but it is fragile, not future-proof, and adds unwarranted complexity.
> 
> Since identityref conformance does not actually require any of the published identities
> to be used at all (only the base has to match), the entire concept of conformance for identities needs
> to be re-evaluated.
> 
> Is it important to restrict the server to specific identities,
> or is it sufficient for the client to simply discover what identities are supported?
> 
> It seems in many cases (e.g. iana-if-type) that the authors clearly do not intend
> whatsoever that every identity in the module be mandatory-to-implement.
> It seems like a discovery mechanism for identityref data nodes would be more
> useful than attempting to prune identities with if-feature-stmts.
> 
> My preference is that all features and if-features related to identities be removed.

If we go the route of moving the identities to IANA-maintained modules, then this will likely happen automatically.   If we go the route of dropping the “common” modules altogether, then this becomes a non-issue.  In either case, the if-featured identities disappear.



> 2) perfect is the enemy of good
> 
> The 5 year milestone is coming up on this work.
> Pencils down. Time to finish.

:sigh:



> Andy

K.