Re: [Netconf] [NETCONF] <get-schema> mandatory parms discussion

[Andy Bierman <andy at netconfcentral.com>]

Mark Scott wrote:
> Andy,
> 
> A rationale for not modifying the mandatory parameters in <get-schema>
> was posted to the ML on Sept 16th.
> 
> I suspect you missed that based on the comment below so I am repeating
> it here for further discussion:
> 
> 	"2.  <get-schema> mandatory parameters:  Initially <identifier>
> was the only mandatory parameter in the <get-schema> operation, however
> after we agreed to expand the key definition in the schema list we made
> <version> and <format> mandatory parameters in the operation.  My
> preference would be to keep this alignment between the key and
> <get-schema> mandatory parms.  I agree in cases where only 1 revision
> and/or 1 format exists this is heavy-handed but it protects against
> agents returning schema a manager may not have intended since
> <identifier> is not unique in schema list and the logic on agent to
> select the default revision may differ from what the manager would have
> selected (esp. if more than 1 version exists)."
> 
> Previously no one else had objected to these being mandatory parameters
> and no ML consensus had been reached to modify it.
> 

I don't understand why the <get-schema> RPC
is coupled to the /netconf-state/schemas data structure.

Revisions are optional in YANG.  A valid module
can exist without a <version>.  This module forces
a YANG module to use the empty string as the
version, which is somewhat of a hack.


> I have changed the subject header of this email in an attempt to
> initiate a ML discussion and will update the draft based on consensus.
> 
> Mark
>

Andy