Re: Naming Conventions for Descriptors (was: comments/review<draft-ietf-ops-mib-review-guidelines-00.txt>)

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

>>>>> C M Heard wrote:
CMH> OK, I have reworked the proposal to use my favorite example
CMH> [which is OPT-IF-MIB (a media-specific extension of IF-MIB)].
CMH> Unfortunately that is still an Internet-Draft (albeit one that
CMH> is in IETF last call).  If you have an equally good example that
CMH> is already published as an RFC let me know and I will use it
CMH> instead.

>>>>> Juergen Schoenwaelder replied:
JS> Equally good will be hard to measure. I personally would
JS> probably use something like the ENTITY-MIB which models
JS> something which is less specific and more generally applicable
JS> (and thus the MIB is probably wider known).

ENTITY-MIB is not suited to the purpose at hand because it does not
have local TCs that follow the suggested naming conventions.  The
TCs defined in it are apparently of a more generic nature, having
names like PhysicalIndex, PhysicalClass, and SnmpEngineIdOrNone.  
OPT-IF-MIB, on the other hand, has many local TCs that follow the
suggested naming convention:  OptIfAcTI, OptIfBitRateK, OptIfDEGM,
OptIfDEGThr, OptIfDirectionality, OptIfSinkOrSource, OptIfExDAPI,
OptIfExSAPI, OptIfIntervalNumber, OptIfTIMDetMode, and OptIfTxTI.

JS> Looking at the OPT-IF-MIB, I actually tend to dislike the module
JS> naming convention. I believe that if should be called IF-OPT-MIB
JS> rather than OPT-IF-MIB since it actually seems to be an
JS> extension of the IF-MIB for optical interfaces. I generally
JS> prefer if the most general part of a module name is first.

That sounds odd to me, and apparently many folks agree ... the only
(published) MIB modules I could find with names of the form IF-* are

IF-MIB
IF-INVERTED-STACK-MIB

while other hand, there are a few of the form *-IF-*

CIRCUIT-IF-MIB
DOCS-IF-MIB
DOT12-IF-MIB

and many others that don't have -IF- in the name but do have the a
prefix indicating the name of the technology (EtherLike-MIB, DS1,
DS3, SONET, and so on).  I think putting the name of the specific
technology first (CIRCUIT-*, DOCS-*, DOT12-*, OPT-*) is a very
defensible convention for the media-specific extentions to the
IF-MIB (it would not have been a good idea for the inverted stack
table MIB module, of course).

Since I am doubtless annoying everyone else with this conversation
perhaps I had better close it off by suggesting that we had better
leave the guidelines vague on the selection of a prefix for a MIB
module name since we apparently can't aways agree among ourselvew
on the best way.

On the issue of using a published MIB module as an example ... the
CIRCUIT-IF-MIB is close, but I'm not so happy with the choice of
ci for the descriptor prefix and Ci for the local TC prefix.  I'm
still open to suggestions, but so far OPT-IF-MIB is the best example
I can find.

//cmh