Re: [Agentx] Some confusion, while watching AgentX-Register-PDU in net-snmp 5.4

[Frank Fock <fock at agentpp.com>]

Hi,

Randy Presuhn wrote:
> > | Note that r.range_subid indicates the (1-based) index of
> > | this sub-identifier within the OID represented by r.subtree,
> > | regardless of whether or not r.subtree is encoded using a
> > | prefix. (See the example below.)

The above is a quote of RFC 2741 page 24.

> > "within the OID represented by r.subtree" tells me that the index is
> > into the *actual* complete OID.  r.subtree is just a representation of
> > the complete OID.
>
> The example shows an r.range_subid value of 5.  For the example
> to make sense, this is the fifth word in the encoding (with a value of 1),
> not the fifth subidentifier of the object identifier it represents.  For your
> interpretation to be correct, r.range_subid in the example would
> be 10.
The example on page 25 and 26 of RFC 2741 reads as follows:

  The use of r.range_subid and r.upper_bound provide a general
  shorthand mechanism for specifying a MIB region. For
  example, if r.subtree is the OID 1.3.6.1.2.1.2.2.1.1.7,
  r.range_subid is 10, and r.upper_bound is 22, the specified
  MIB region can be denoted 1.3.6.1.2.1.2.2.1.[1-22].7.
  Registering this region is equivalent to registering the
  union of subtrees

  1.3.6.1.2.1.2.2.1.1.7
  1.3.6.1.2.1.2.2.1.2.7
  1.3.6.1.2.1.2.2.1.3.7
  ...
  1.3.6.1.2.1.2.2.1.22.7

Following that example, I would say that 14 is the correct value
for r.range_subid in Mark's original example.

> Note that the text says "r.subtree", *not* "r.region", and thus is not
> in terms of the complete object identifier that r.region encodes.
> At least that's how I read it. 
>
> Either the example is incorrect, or the explanatory text is unclear, or both.  :-)
I think, this has been clarified in RFC 2741 which obsoletes 2257 ;-)

> > I supposed it could be more clear.  The proof of the pudding would be
> > an interop-fest, I supposed.

I have implemented AgentX++ and SNMP4J-AgentX according to
RFC2741 and so far got no complains about interoperability
issues - except those I had with the NET-SNMP master and
sub-agents regarding handling of search-ranges and region
registration. The bug(s) Mark found might be one reason
for that.

Regards,
Frank Fock