Re: Adjacency index

saperia@tcpjon.ogo.dec.com Fri, 28 August 1992 16:36 UTC

Received: from ietf.nri.reston.va.us by IETF.NRI.Reston.VA.US id aa04159; 28 Aug 92 12:36 EDT
Received: from NRI.NRI.Reston.Va.US by IETF.NRI.Reston.VA.US id aa04155; 28 Aug 92 12:36 EDT
Received: from inet-gw-2.pa.dec.com by NRI.Reston.VA.US id aa10633; 28 Aug 92 12:38 EDT
Received: by inet-gw-2.pa.dec.com; id AA06347; Fri, 28 Aug 92 09:37:19 -0700
Received: by nsl.pa.dec.com; id AA10059; Fri, 28 Aug 92 09:18:03 -0700
Received: by nsl.pa.dec.com; id AA10055; Fri, 28 Aug 92 09:18:02 -0700
Received: by inet-gw-2.pa.dec.com; id AA05356; Fri, 28 Aug 92 09:18:01 -0700
Received: by tcpjon.ogo.dec.com (5.57/ULTRIX-fma-071891); id AA10842; Fri, 28 Aug 92 12:20:32 -0400
Message-Id: <9208281620.AA10842@tcpjon.ogo.dec.com>
To: Bob Stewart <rlstewart@eng.xyplex.com>
Cc: saperia@tcpjon.ogo.dec.com, phiv-mib@pa.dec.com, davin@thumper.bellcore.com
Subject: Re: Adjacency index
In-Reply-To: Your message of "Fri, 28 Aug 92 11:07:56 CDT." <9208281607.AA12928@xap.xyplex.com>
Date: Fri, 28 Aug 92 12:20:31 -0400
From: saperia@tcpjon.ogo.dec.com
X-Mts: smtp

Bob,

No problem, Chuck sent me a note which echo's what you say:


>Jon,
>
>This question is easy. OIDS are NEVER EVER EVER reassigned and used
>for another purpose. So, the current OIDs must continue to refer
>(forever) to the table that you are going to obsolete. You must
>allocate new OIDs to name the objects in the new-and-improved version
>of this table. I would suggest that you define the adjacency group at
>one place in the MIB document.  It should include two tables: the old
>one (for which the names of the objects are the old OIDs and for which
>the status of the objects is "obsolete") and the new table (for which
>the names of the objects are all new OIDs and the status of the
>objects is "mandatory").

So I will do what he said, keep existing group intact with table intact.  Add
new table to the group, old table and all objects in it will be obsolete.

With regard to the sematics of the CircuitName/Index what do you think about
using CircuitIndex.  It is a simple integer making the index simple to code
(along with the NodeAddr) - you do loose the ability to get at all the
adjacency informaton on a single circuit this way.

John Shriver said:

>>I got the impression that a CircuitName, which you were proposing as
>>an index, was a string of some sort.  (As compared to CircuitIndex,
>>which is a number.)

He is correct Name would be a string (yes it really does not matter if we call
it a DisplayString or an OCTET STRING).  I would still be of a variable lenght.

I also just got a note from Art which said:

>But please, let's use phivAdjCircuitIndex, it's much simpler to deal with.

OK lets use phivAdjCircuitIndex The DESCRIPTION of this would be:

  	       "A unique index value for each known circuit.  This
               value is the same as phivCircuitIndex and identifies the
               circuit over which the adjacency is realized."

Is this what everyone expects??

I am going around on this so that everyone is comfortable with the change and
understands what it means (including myself).