RE: draft-ietf-ops-mib-review-guidelines-01.txt is now available

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

On Tue, 25 Feb 2003, Andrew Smith wrote:
> I make particular objection to Andy's statement that these are
> "practical guidelines on how to configure these programs to
> check a MIB so it will meet IETF expectations". This is newspeak
> for "if you don't buy or otherwise procure these products, you
> won't get your MIB past the MIB quacks and/or IESG members".

I'll let other folks speak for themselves, but that most assuredly
will not be the case if I am assigned as your MIB reviewer.  On more
than one occasion in the past I've has the job of reviewing a MIB
module written by someone without access to a MIB compiler and I
have in those cases spent a great deal of time and effort telling
the authors how to fix the compilation problems.  If necessary, I
will do that again.  However, the process invariably goes a whole
lot faster -- both for me and for the author -- if the stuff I get
from the author will at least compile cleanly.  Several times in the
past I've sent instructions like those in Appendix B and Appendix C
of the draft to authors who wanted to know what tools to use and how
to use the tools effectively.  So it seemed reasonable to include
that stuff in the document.

> It also implies that the standard is not sufficiently
> documented.  I don't think IETF should be making such
> implications (the latter, if true, needs fixing).

When I run a MIB module that I am reviewing through _any_ kind of
MIB compiler, I turn on all of the warnings that I think could be
meaningful, but I _do not_ assume that a diagnostic from the tool
invariably indicates a problem.  I _always_ check the underlying
standards documents (in this case, RFCs 2578, 2579, and 2580) to
make sure.  The reason for using the tool (rather than going through
the module by hand) is that it will have a high probably of finding
stuff that I might well miss.  Under no circumstances do I ever
substitute the judgement of the tool for my own.  I agree completely
with the following statement of Andy Bierman's:

% SMI conformance is dictated by the standards documents, not any
% particular SMI conformance test tool.

Thanks,

Mike