Re: Guidelines Section 4.6.5 (OID Length Limitations and TableIndexing)

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

On Wed, 5 Feb 2003, Michael Kirkham wrote:
> | 4.6.5.  OID Length Limitations and Table Indexing
> 
> It might also be a good idea in this section to recommend against using
> objects of type OBJECT IDENTIFIER as indices since they cannot be subtyped
> under any circumstances.

I'm definitely not going there.  It would be another CLR.  (I got
beat up for that once before and I don't want it to happen again.)

> Perhaps also a DESCRIPTION for such a table should (must?) mention unified
> limits where it's not possible to represent the limits syntactically or
> where limits aren't present (due to a desire to not impose an arbitrary
> limit on one of multiple indices where it's the combination of all of the
> indices for a particular row that have the hard limit).

That's been suggested here before (by me, I think).  Note that
it can happen with OCTET STRING as well as OBJECT IDENTIFIER
(the APM-MIB has tables with several OCTET STRING indices where
the only meaningful limitation was is aggregate one).

Do we need guideline text for that?

//cmh