Re: dsx1/dsx3CircuitIdentifier still deprecated?

[Gary Hanson <[e-mail]>]

Dave,

As far as I can remember, the two objects were planned to be deprecated
back at a time when it seemed possible to increase the MAX-ACCESS of
ifPhysAddress to be read-write, at which point the ifPhysAddress object
could be used instead of the dsx1/dsx3CircuitIdentifier objects.

However, the conclusion of the ifmib WG (from the August, 1996 minutes)
was that there was going to be no change of ifPhysAddress, and indeed
no change has been made to the draft-ietf-ifmib-mib-05.txt document.

So, at that point, the dsx1/dsx3CircuitIdentifier objects should have
been restored to current status.  What complicates thing for us now
is that dsx3CircuitIdentifier was never flagged in the ASN.1 of the
draft-ietf-trunkmib-ds3-mib-05.txt document as deprecated, even though
the list of changes and the discussions of the use of ifPhysAddress in
that document document SAYS it was.

Given that the DS3-MIB draft is thus inconsistent with itself at best,
and inconsistent with the ifmib draft at worst, I think it makes sense
to restore the objects's MAX-ACCESS back to current status, since the
users of these MIBs have to have SOME object where the values can be
written.  (Note that it could remain an implementation choice whether
or not to populate the ifPhysAddress with these values, once written.)

I apologize for not noticing the inconsistencies earlier, but I think
it needs to be fixed.

Regards,
Gary


On Fri, 1 Aug 1997, David Fowler wrote:

> Dawn Xie wrote> 
> > CircuitIdentifier object is useful to associate a port to a 
> > connection facility. So I agree with Gary that these objects better 
> > not be deprecated.
> > 
> 
> Gary Hanson wrote: 
> > I thought a conclusion had been reached a year ago about not 
> expanding
> > the MAX-ACCESS of ifPhysAddress, and the current IF-MIB 
> draft (which has
> > apparently expired since May!) still shows ifPhysAddress to be 
> read-only.
> 
> > > 
> > > Do you know which direction this discrepancy is going to be 
> > resolved?
> > > 
> 
> The last drafts I submitted are intended to be the final documents.
> This means that CircuitIdentifier was supposed to stay deprecated.
> 
> I, for one, can see the value of naming objects.  I guess the question
> is whether one truly needs to set these objects through SNMP or will they 
> be assigned on a local interface and then only read through SNMP?
> 
> Gary/Dawn:  Will you give me a bit more info on how you would like to
> use this object?
> 
> Thanks,
> David
> 
> P.S.  No promises here.  These drafts have been merely waiting for
> interoperability and I'm not planning on churning them again unless
> the chair asks me to do it!
>