RE: Index values of zero (was: Index values of zero)

["C. M. Heard" <heard@pobox.com>]

On Wed, 22 Jan 2003, Wijnen, Bert (Bert) wrote:
> Fine with me, but I thought that some of you had stated that
> by adding that SHOULD [describe] it in DESCRIPTION clause was
> adding another CLR. That was not my intention, there for I
> suggested: "It is good practice to.... " which still gets the
> message across and does not sound like another CLR.

The original language was "[if] 0 is included in the range, then
a good reason MUST be specified."  Keith McCloghrie, in the e-mail
that you forwarded to us, objected thusly:

  As for 0, I think "a good reason MUST be specified" is bogus; where
  must it be specified ?  and who decides whether it's good or not ??

and that went away when I re-worked the text to include the situations
in which he suggested use of zero was reasonable.

The way I see it is that we would get the same thing in practice
-- i.e., a rule that a mib doctor would be expected to enforce --
whether or not the capitalized SHOULD is used in the replacement
language, so I figured that we might as well be honest and put
it there explicitly.  And it's probably justified:  when an index
variable uses a value of zero to indicate some kind of special
case, then that usage does in fact need to be spelled out in the
DESCRIPTION clause so that an implementor will know what to do.
A good example of what I think we should expect is provided by
the ifStackHigherLayer and ifStackLowerLayer index columns in the
ifStackTable of RFC 2863:  they use the InterfaceIndexOrZero TC and
state that zero means no higher (respectively lower) layer exists.

//cmh