Re: Doubts about rfc 1447

cyoung@ccmailpc.ctron.com Mon, 02 August 1993 21:27 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa12331; 2 Aug 93 17:27 EDT
Received: from CNRI.RESTON.VA.US by IETF.CNRI.Reston.VA.US id aa12327; 2 Aug 93 17:27 EDT
Received: from SLEEPY.TIS.COM by CNRI.Reston.VA.US id aa27745; 2 Aug 93 17:27 EDT
Received: from sleepy.tis.com by sleepy.TIS.COM id aa02764; 2 Aug 93 21:08 GMT
Received: from tis.com by sleepy.TIS.COM id aa02762; 2 Aug 93 17:01 EDT
Received: from thumper.bellcore.com by TIS.COM (4.1/SUN-5.64) id AA14687; Mon, 2 Aug 93 17:01:57 EDT
Received: from nic.near.net by thumper.bellcore.com (4.1/4.7) id <AA25160> for snmpv2@tis.com; Mon, 2 Aug 93 17:01:54 EDT
Received: from ctron.com by nic.near.net id aa01508; 2 Aug 93 17:00 EDT
Received: from stealth.ctron.com by ctron.com (4.1/SMI-4.1) id AA16229; Mon, 2 Aug 93 17:00:28 EDT
Received: from express.ctron.com by stealth.ctron.com (4.1/SMI-4.1) id AA03028; Mon, 2 Aug 93 17:00:22 EDT
Received: from ccmailpc.ctron.com by express.ctron.com (4.1/SMI-4.1) id AA19598; Mon, 2 Aug 93 16:55:51 EDT
Received: from cc:Mail by ccmailpc.ctron.com id AA744335300 Mon, 02 Aug 93 16:48:20 EST
Date: Mon, 02 Aug 93 16:48:20 EST
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: cyoung@ccmailpc.ctron.com
Encoding: 2454 Text
Message-Id: <9307027443.AA744335300@ccmailpc.ctron.com>
To: snmp@psi.com, F.Goncalves@cs.ucl.ac.uk
Cc: snmp2@thumper.bellcore.com
Subject: Re: Doubts about rfc 1447


>o in order to consider a row from the aclTable as ready for activation,
> do the parties/context referred in it have to exist? And what about
> the contextTable?

I would say yes to both.  Though you are right in pointing out that no
procedure is documented what happens to an ACL entry when a party or
context is deleted.

>o rfc 1447 says (pg 33) about contextProxyDstParty:
>  "If the value of the corresponding instance of the
>  contextViewIndex is greater than zero, then the
>         value of an instance of this object is { 0 0 }."
>
> This means that:
>  - is {0 0} returned for get requests ?

Yes.

>  - is contextProxyDstParty automatically set to {0 0} ?

not necessarily.  This is not explicity said.

>  - is the set operation on contextViewIndex refused unless 
>    contextProxyDstParty is {0 0} ?

No.  This is not implied in the RFC.
  
>o rfc 1447 says (pg 35) about contextStatus:
>  "A context is not qualified for activation until
>          instances of all corresponding columns have the
>          appropriate value.  In  particular, if the
>          context's contextViewIndex is greater than zero,
>          then the viewStatus column of the associated
>          conceptual row(s) in the viewTable must have the
>          value `active'.  ..."

> Should this be checked when setting contextViewIndex?
> I doesn't seem to be mandatory to do it but, it could be useful
> for an user. 

No.  A row in the not-ready status can have bogus information in it
legally.  If you decide to only allow active viewIndices for the
contextViewIndex then I do not think you will break anything either for
that manner.  It seems this can be implementation specific withouth blowing
anything up.

> In general, for managed objects whose value has some constraint like
> that (another example is the size of partyTAddress when partyTDomain
> is { snmpUDPDomain }), should the values provided in set requests
> be checked or this should be delayed until activating
> (setting RowStatus to active / notInService) the row?

As above, either way could be legal.


>Please state in your answer if it corresponds to something mandatory or 
>is a recommendation based on your experience.


I think only one of my answers above (the "Yes." one) is hard in stone.
The others I have tried to answer in the "spirit" of the model but I'm 
sure it could be argued another way.

Chris