Re: WG last call: Coex draft

On Mon, 30 Dec 2002, Mark Ellison wrote:
> "C. M. Heard" wrote:
> > It's true that rfc 2576 and the current coex draft don't handle
> > notifications with Counter64 variables in them in the same way
> > as did rfc 2089, but that was a deliberate change.
...
> What I'm unable to find is a trail of this deliberate change.
[ ... ]
> [I]n coex-04 (~may-21-1999) the language in relevant section (3.2)
> was changed to specify the notification cannot be sent using SNMPv1.
> 
> I'm unable to find any thread in my SNMPv3 email archive discussing the
> change reflected in coex-04, nor did I see anything in the minutes from
> Orlando or Minnesota WG meetings discussing the change reflected in
> coex-04.

It's well-hidden, but it's in the IETF March 1999 Proceedings at the
SNMPv3 WG minutes for the IETF-44 meeting in Minneapolis.  It says:

   [ ... ] we decided to follow standard not-in-view procedure and
   drop notifications going out as SNMPv1 but coming in with Counter64.

Notice that this is in line with what would happen in a pure SNMPv3
environment:  a notification is not sent to a given target if any
of the mandatory varbinds would be out-of-view for that target.  There
is a good reason for doing this:  a notification definition (whether
specified by an SMIv2 NOTIFICATION-TYPE or an SMIv1 TRAP-TYPE
definition) lists mandatory varbinds that a notification receiver
can always expect to receive.  Selectively stripping away out-of-view
varbinds as specified in rfc 2089 would violate that contract.

//cmh