RE: Re: Draft-iet-bridge-bridgemib-smiv2-09.txt

Hi Juergen,

Personally, I thought the original vague text was fine, as was the
original reference to RFC1213.
I agree that most implementations would simply do all seven system
objects, and similarly for interfaces.

But we have competing interests. You and Bert want to get rid of
references to RFC1213. Mike wants to get rid of inaccurate statements
about relationships to other mibs. I would like to keep the RFC1213
reference because compatibility with RFC1493 has been the primary
goal, but unless others in the WG speak up and say that such
compliance is the consensus and that RFC1213 is fine, as co-chair, I
need to try to adjudicate the competing consensuses.

One consensus is that we shouldn't use a reference to RFC1213. This is
the view of you and Bert; we have two persons, one being an O&M AD.
Mike and I are willing to accept that using it is undesirable, but not
critical.
Mike tells us that the OPS AD insists on accurate cross-document
dependencies; again, we have two persons, one being an AD. (We haven't
heard David actually say this however). 

Unfortunately, these two views are hard to reconcile; if you point to
RFC3418 and RFC2863 instead of RFC1213, the statements become
inaccurate; if you point to the available confirmance clauses, the
document becomes inaccurate; if you list all seven system objects and
the many interfaces objects as dependencies, the document becomes
inaccurate.

Me, I want see the document finished so we can move on to other
documents in the WG charter, and move toward WG closure. What I see is
a debate between two entrenched positions, neither of which helps an
operator manage a device. This debate is all about IETF publication
process and neither way of specifying the relationship between mib
modules offers anything substantive to an operator.

Will you yield on your insistence to not use RFC1213? Bert, will you
yield on this point? If so, consensus could move to "while we are
referenceing RFC1213, and that may be undesirable, using RFC1213 in
the statement of relationships between mib modules is in fact the
closest we can get to accurately portraying the original intent, and
therefore its usage should be acceptable."

David Harrington
[email protected]
co-chair, IETF Bridge WG
 

> -----Original Message-----
> From: Juergen Schoenwaelder [mailto:[email protected]] 
> Sent: Friday, January 28, 2005 2:49 AM
> To: David B Harrington
> Cc: 'Wijnen, Bert (Bert)'; 'C. M. Heard'; 'Bridge-Mib (E-mail)'
> Subject: Re: [Bridge-mib] Re:
Draft-iet-bridge-bridgemib-smiv2-09.txt
> 
> On Thu, Jan 27, 2005 at 08:34:14PM -0500, David B Harrington wrote:
> 
> [...]
>  
> > --- comments on Juergen's proposed text:
> > Section 3.2.1: I believe Juergen's proposed text that says
> > 'Implementation of the BRIDGE-MIB module requires
> > that at least ... are implemented' is inaccurate; this mib 
> module can
> > be implemented without those objects if desired, but you will miss
> > some of the objects recommended by 802.1-D, and the 
> functionality they
> > provide (but then even if you implement this mib module plus those
> > objects, you'll still be missing some objects listed in 802.1-D
and
> > the functionality they provide, as described in 3.1).
> 
> Obviously there is no common view what RFC 1493 was trying to say in
> section 3.2.* and this is why I orginally liked to keep some rather
> vague text. In practical terms, I have never seen a bridge which did
> not implement the seven system scalars so from a practical 
> perspective,
> this whole debate is really a mood point and a waste of time.
>  
> > Section 3.2.2: I believe Juergen's proposed ext that says
> > 'Implementation of the BRIDGE-MIB module requires that the
following
> > objects of the IF-MIB module are implemented' is inaccurate in its
> > list, as I describe above.
> 
> It boils down to how you read the first paragraph in 3.2.2. 
> My reading 
> is that it says segments are considered subnetworks from the 
> interfaces 
> point of view and that the interfaces group is mandatory for 
> all systems. 
> Again, I have never seen a bridge which did not implement the full 
> interfaces group.
> 
> /js
> 
> -- 
> Juergen Schoenwaelder		    International University Bremen
> <http://www.eecs.iu-bremen.de/>	    P.O. Box 750 561, 
> 28725 Bremen, Germany
>