Re: [NGO] external module properties

[Juergen Schoenwaelder <j.schoenwaelder at jacobs-university.de>]

On Sat, Apr 26, 2008 at 02:38:52PM -0700, Andy Bierman wrote:
> Juergen Schoenwaelder wrote:
> > On Sat, Apr 26, 2008 at 11:28:20AM -0700, Andy Bierman wrote:
> >  
> >> LAST-UPDATED and REVISION are optional in SMIv2.
> > 
> > This is not correct. The SMIv2 does mandate a LAST-REVISION clause
> > (see section 5 and 5.1 of RFC 2578).  The IETF guidelines in addition
> > mandate the presence of REVISION clauses and the LAST-UPDATED clause
> > becomes redundant when you have REVISION clauses. That is why YANG
> > only has revision statements.
> 
> some modules like SNMPv2-TC have no MODULE-IDENTITY section

The modules SNMPv2-SMI, SNMPv2-TC, SNMPv2-CONF defintion the language
itself and are not proper MIB modules anyway. And in any case, a
counter example would not invalidate what is written down in RFC 2578.
 
> > The YANG language itself does not mandate the presence of one or more
> > revision statements. This is consistent with YANG's philosophy of
> > mandating only those things that are essential for the language to
> > function and leave other things to guidelines or applicability
> > definitions.
> 
> I think the NETMOD philosophy should consider the entire
> network configuration problem, which includes properties like
> module version.

It does. There is a revision statement and all the other stuff needed
for the entire network configuration problem.

> > PS: If YANG were to support versioned imports, revision clauses
> >     would have to be mandatory because without version information,
> >     a compiler would not be able to resolve imports. Right now,
> >     compilers can do the right thing without having revision
> >     statements and hence they are optional.
> 
> External entities (humans, applications) need to be able to distinguish
> between multiple versions of the same module.  Why does a YANG module
> need to include a 'namespace' clause?  Maybe some vendors want to
> ignore namespaces.  IMO, it should not be optional to provide a version
> identifier.  This is a separate issue from requiring the version identifier
> to change under certain conditions.

Namespace information is needed for parsers to work properly in case
of identifier clashes. And I am sure the IETF will require that
revision clauses are present in YANG modules. But this is not a
technical requirement of the language itself; it is a usage guideline
requirement. I believe this distinction is worthwhile to make.

/js

-- 
Juergen Schoenwaelder           Jacobs University Bremen gGmbH
Phone: +49 421 200 3587         Campus Ring 1, 28759 Bremen, Germany
Fax:   +49 421 200 3103         <http://www.jacobs-university.de/>
_______________________________________________
NGO mailing list
NGO at ietf.org
https://www.ietf.org/mailman/listinfo/ngo