Re: Mapping of the MODULE-IDENTITY value

On Tue, 3 Dec 2002, David Spakes wrote:

> Why couldn't the OBJECT IDENTIFIER for "snmp" have been reused in a
> MODULE-IDENTITY macro?
>
> What difference is there between invoking the MODULE-IDENTITY and defining
> a non-leaf OBJECT IDENTIFIER?
>
> Consider the mosy tool.  When run with the argument "-1", mosy converts
> an SMIv2 document to SMIv1.  In the process, it maps MODULE-IDENTITY to
> OBJECT IDENTIFIER (see example below).  Why wouldn't it be okay to go
> the other way around when one is republishing a MIB document in SMIv2?

I'm also curious what others think about this.  MIB Smithy's converter,
actually does go ahead and convert the top-most OBJECT IDENTIFIER in the
module to a MODULE-IDENTITY, but only if there is just one.  If there are
multiple choices, then it's left to the user to create a MODULE-IDENTITY.

I suppose one reason it might not be legal is that multiple modules may
define the same node locally, perhaps with different labels, which is
legal (at least in ASN.1)... But there is the potential that both become
conflicting MODULE-IDENTITY invocations if both are converted and both had
the same top-most node.  Seems like a pretty big stretch, however.  How
many modules do you know of that define the same OBJECT IDENTIFIERs where
it isn't due to one being made obsolete by the other?  I don't think I
know of any, personally.

--
Michael Kirkham
www.muonics.com