[SNMP4J] Basic Question about implementing SNMP Tables

[Frank Fock]

inliner683 at gmx.de wrote:
> 
> Is is not allowed to create a "new" MIB in SMIv1?

Yes, it is not allowed (by the IETF) to create
a new MIB with SMIv1.

> Should it not be a design decision which version to 
> use (of course ignoring all benefits from SMIv2 when
> using SMIv1)?
>
No, because SMIv1 has some flaws (one you discovered
here).

>> and since it can be easily implemented by the
>> instrumentation.
> 
> That's true. Nevertheless the process I initially expected 
> for creating a row was:
> 
> 1) setting all index columns with appropriate values

Normally, there are no index column objects. May be
that's the misunderstanding. Objects with maximum
access "not-accessible" are virtual objects only,
that in almost all cases are not implemented as
managed objects in the agent.

> 2) setting all other columns with desired values
> 3) creating a row with the column information altogether
> 4) automatically getting rows with appropriate OIDs (with no requirement to set the OID anymore, since
>    OID is known from index columns)

See above, the other way around makes sense.
Create the index from its sub-index object values and
only set the accessible column objects.

> 
> In SMIv2 the only difference is that the index columns are not accessible via e.g. a MIB browser, but the
> value of the index columns must be defined as well (currently implicitly done via the OID parameter in 
> createRow()-method).

Here is the misunderstanding. The not-accessible columns
simply do not exists.

Best regards,
Frank

-- 
AGENT++
http://www.agentpp.com
http://www.mibexplorer.com
http://www.mibdesigner.com