Re: [netmod] Wrapping up system-creatable

[Phil Shafer <phil at juniper.net>]

David Partain writes:
>1. Do nothing in the draft and let implementations and data models figure out
>how/if/when/why the server can modify the configuration data.

-1

>2. Do nothing but write in the document that we _know_ it's an issue but we're 
>still not doing anything.

This is the minimum we need to do.

>3. Add text that says that the server MUST NOT modify config data unless 
>dictated by the data model (through description clauses or future 
>extensions).

-1.  "MUST unless you make an extension" has no force.

>4. Add text that says that the server is free to modify config data in any way 
>it wants, unless constrained by the data model (through description clauses 
>or future extensions).

ditto.

>5.  Add formal statements (e.g. the current s-c) to control how the server can 
>modfiy config.

I'd learn toward this, but I really think there's too much
conflict to move down this path.  s-c should be broken out
into a distinct I-D.

So I guess that makes me a "2".

Thanks,
 Phil