Re: [NGO] NETCONF Data Modeling BoF (NDM) proposal

Balazs Lengyel wrote:
> Hello Andy,
> You yourself sad that XSD is
> - write-only
> - prone to errors
> - you yourself choose not to use it
> - is like programming in assembler
> 
> (Which I fully agree with.)
> 
> I am surprised that you still want it as the main modeling language for 
> Netconf.

I don't really see any better alternatives.
One can simplify the problem by drastically limiting
what one can do within XML, within NETCONF, as a way
to address the problem.  But the NETCONF WG has already
used many complex features of XSD, such as substitutionGroups
and choices, because they were needed to express the data model complexity.

I think a new DML is fine for a background experiment,
not as something in the critical path for producing
any standard data models.

The advancement of protocol interoperability can only helped
by standardizing Xpath object/instance identifiers for better
<error-path> usability, or defining how the continue-on-error
option actually works, or defining which top level element is used
for the target of a <copy-config> operation, or dozens of other
details that implementers are aware of...

> 
> The fact that IETF could not agree on SMIng should not just mean that we 
> avoid modeling in the future, but also that we accept, that there is a 
> stronger need for compromise.
> 
> As I see it, a  good part of SMIng's problem were due to compatibility 
> constraints with SNMP, which is less of a problem for a Netconf modeling 
> language.

SMIng had many problems, including COPS-PR integration as part of the charter.
There is not going to be consensus on another SMIng without producing
a requirements document, conducting a candidate evaluation phase, and
then a standard definition phase.  That takes 5 years in the IETF, if you are lucky.

What about the tools problem?
The whole point of using XML was to take advantage of mainstream/free tools
and not repeat the SMIv2 mistake (adapted subset of obsolete version of ASN.1)
of being the only ones using the DML.

> 
> Balazs
> 
> 
> 

Andy

_______________________________________________
NGO mailing list
NGO@ietf.org
https://www1.ietf.org/mailman/listinfo/ngo