Re: FW: Minutes from the sming wg interim meeting, June6-8,2002 in Washington DC

[Steve Waldbusser <waldbusser@nextbeacon.com>]

David,

  You're right. I think it really reflects an error in the discussion
and/or minutes rather than an error in the solution.

  You'll be able to skip any zeros in the base during your traversal to
the top-level table node. For example, if fooTable is { 0 1 0 1 0 1 },
then you'll skip the zeros as you parse down to figure out that it's in
the fooTable. At that point the next node cannot be zero (fooEntry) nor
can nodes below that.

Another way to put it is that the zero 'separator' must be after the
fooTable node. You shouldn't be looking for the separator until after
you've found the top-level table.

But we'd better be sure we outlaw zeros in the hierarchy in SMI-DS just
as we have in 2578 (I think Andy has already done this).


Steve


"David T. Perkins" wrote:

> HI,
>
> Note guys that there is a technical assumption in the following that
> is not correct. That is that a zero valued sub-identifier cannot
> occur in the OID for an object type. The only restriction that the
> SMI has it that the last sub-id cannot have a value of zero. Thus,
> any sub-id from the root (and even the root) can have a value of
> zero.  For example, it would be valid to have the following
> definition for an object type:
>
>    foo OBJEC-TYPE
>        ...
>        ::= { 0 1 0 1 0 1 }
>
> The rules are in RFC 2578, section 7.10 (last paragraph).
>
> More comments later...
>
> Regards,
> /david t. perkins