RE: draft-ops-endpoint-mib-04.txt

[Andrew Smith <andrew@extremenetworks.com>]

I think I was one of the ones who persuaded the authors to tighten up the
language to say "MUST NOT". I don't actually care whether it says "MUST" or
"MUST NOT" but I do care about removing implementation options if they are
gratuitous: my line of reasoning was based purely on promoting
interoperability.

On your second point though, if there are multiple addresses in an INDEX
clause then you cannot rely on the number of OIDs remaining. But you might
be able to rely on knowledge of the InetAddressType to implicitly know the
length. 

Andrew

> -----Original Message-----
> From: Bill Fenner [mailto:fenner@research.att.com]
> Sent: Tuesday, January 11, 2000 11:27 AM
> To: mibs@ops.ietf.org
> Subject: Re: draft-ops-endpoint-mib-04.txt
> 
> 
> 
> >   The IMPLIED keyword MUST NOT be used for an object of type
> >   InetAddress in an the INDEX clause.
> 
> Any particular reason for this?  Since there's an InetAddressType
> earlier in the index, there aren't any ordering issues that I can
> see that would be helped by having the length field, and the length
> of the InetAddress is obvious from the number of OIDs remaining.
> 
> Thanks,
>   Bill
>