HI,

I agree with Wes below on the technical point that requiring
support for "createAndWait" is upto the MIB designer and
specifier of the MODULE-COMPLIANCE specification.
However, I disagreee with his label of "cheaper agents" in
characterization of why only "createAndGo" is specified
and used by some MIB designers and agent implementors.

For example, I believe that "createAndWait" was just a hack,
and should never be used in MIB design. Likewise, "notInService"
is another hack that never be used.
I believe that tables that use only "active", "createAndGo",
and "destroy" are easy to understand, implement, and use.


On Mon, 12 Jun 2006, Wes Hardaker wrote:
> >>>>> On Sat, 03 Jun 2006 09:18:32 +0000, "Anurag Verma" <[EMAIL PROTECTED]> 
> >>>>> said:
> 
> Anurag> So, from subagent implementation, I was thinking that for this
> Anurag> to happen, the SNMP subagent should "cache" all the requests
> Anurag> coming in different SNMP PDU and create a row instance for the
> Anurag> same and populte that row instance as when different PDU come.
> Anurag> So, my questions are:
> 
> Basically, there is some code to help you with this if you use the MFD
> handler type (I believe).  However, the data is cached not by the main
> agent by in a structure that is later turned on.  Robert Story would
> actually know more about this...
> 
> One comment though: supporting createAndWait is actually optional.  If
> you read the RowStatus TC carefully you'll find that agents don't have
> to implement both forms.  In fact, it's fairly common for cheaper
> agents to only implement support for createAndGo.
> -- 
> Wes Hardaker
> Sparta, Inc.
Regards,
/david t. perkins



_______________________________________________
Net-snmp-coders mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/net-snmp-coders
  • Re: none David T. Perkins

Reply via email to