RE: DateAndTime TC oversight in RFC 2579

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

On Mon, 22 Dec 2003, Romascanu, Dan (Dan) wrote:
> Can you clarify why the proposed change would contradict RFC
> 2578?

I don't think that it would, _if_ it was carefully drafted.  In
particular, for any object definition that uses the null value, the
DESCRIPTION clause MUST specify the the circumstances under which
that value can be returned.  Just as with zero in
InterfaceIndexorZero.

> IMO, it looks like
> 
> a. it does not cause interoperability problems over the wire -
> as the MIBs that do not care about the default value need not
> use the clarification, and the other already do it, as proved by
> the list forwarded by Bert

And that brings up the second "careful drafting" requirement:
the default (if not otherwise specified its DESCRIPTION clause)
is that an object does NOT return a null value, and that it's
not legal to modify an object definition later on to make it do
so (changes values managers may get, so is a semantic change).

> b. The change may be implemented using an editorial procedure
> allowed by RFC 2578:
> 
> '(8)  Clarifications and additional information may be included in the
>      DESCRIPTION clause.'

Sure.  But that should be done by publishing an RFC that amends RFC
2578, and not by posting it in the RFC Editor errata list (which is
supposed to be for typos, not technical changes).

Thanks,

Mike