RE: MIB module root assignment

["C. M. Heard" <heard@pobox.com>]

On Fri, 19 Sep 2003, Wijnen, Bert (Bert) wrote:
> Glenn Waters wrote:

> > Could we have a "developmental" branch and place under that a
> > unique number. IANA could be the authority to hand out the
> > unique number under developmental. The IANA process would have
> > to be real simple and they would have to pretty much say yes to
> > anyone.
> > 
> Basically we do have such a thing, and it is called experimental.
> But when people then move from experimental to mib-2, do we then
> require them to rename their modules and descriptors?

There is no formal requirement in the SMI to rename modules and
descriptors when moving from the experimental subtree to the mgmt
subtree (of which mib-2 is a sub-branch).  Uniqueness requirements
for modules names and descriptors are levied only on "standard" MIB
modules (RFC 2578 Sections 3 and 3.1), and the mgmt subtree is used
to identify objects in such MIB modules (RFC 2578 Section 4.1).

> Another thought is that we just use a fixed "nnn" or "xxx" value
> and that MIB compilers will NOT flag an Error, just a Warning
> that that number needs to still be assigned. Other then that
> they would just compiles as if it was a valid number.

I DON'T like that idea.  The fact that "nnn" or "xxx" will not
compile is a feature, not a bug (as David Harrington and I keep
insisting).  In any case, this could only apply to IETF MIB review
syntax checkers;  there is no way you'd want to do it with a
production MIB compiler used to generate agent code or to load a MIB
module into a management application.

//cmh