Re: [NGO] NETMOD charter

[Balazs Lengyel <[e-mail]>]

I agree with Bert. For me this also means we do not need to debate now whether YANG will be 
mapped to DSDL or XSD.  Clearly the first step is to standardize YANG (the operator friendly 
language). Once we have that, if there is enough interest, a mapping to XSD/DSDL will be 
standardized. The two do not exclude each other!!! If there is interest both can be done.


- first step: define what we are mapping (YANG)
- second step: once you have something to map (YANG) decide where and how to map it.

Balazs

Bert Wijnen - IETF wrote:
> Andy writes (in below email):
>> In practical terms, it seems that document reviewers
>> will have a lot more work to do, verifying that:
>>
>>    - all these mappings are well-formed
>>    - they match the semantics of the high-level model
>>    - there are no conflicts between any of the machine-mappings
>>    - all DM variants in the document describe the exact same set of XML
>>
> The way I had understood our discussion last week is a bit
> less heavy. My understanding was:
> 
> - WG develops a User/Operator friendly readbale DML (YANG would
>   be our starting point, so I will use that for now).
>   This will be a stds track doc.
> - WG defines mapping rules from YANG to XML (for the on the wire
>   encoding). This will be a stds track doc.
> - WG defines mapping rules from YANG to XSD (and/or RELAXNG or DSDL).
>   This will be a (or more) stds track document(s)
> - When WGs write a datamodel, they just include the YANG-module(s)
>   as normative.
> - All the other derivations can be generated by (multiple and
>   possibly competing) tools based on the stds track mapping rules.
>   The tools themselves would be out of scope of the WG (similar
>   to tools like SMICng, smilint, ASN.1 verifier; ABNF validation,
>   XSD and XML validation tools).
> - Any additional tools (like the example XSD to graphical representation
>   tool that Sharon showed; but also many others) are out of scope
>   of this WG.
> 
> So that would mean we have some extra work now (namely specing out
> all the mapping rules). But once the DML and the mapping rules have
> been published, the actual Data Model documents themselves can be
> straight forward and simpell with just a YANG module.
> 
> We can (in my view) also decide to do just one mapping rule document
> in our initial release.
> 
> Bert Wijnen
> 
>> -----Oorspronkelijk bericht-----
>> Van: [e-mail] [mailto:[e-mail]]Namens Andy
>> Bierman
>> Verzonden: maandag 17 maart 2008 16:12
>> Aan: Romascanu, Dan (Dan)
>> CC: NETCONF Goes On
>> Onderwerp: Re: [NGO] NETMOD charter
>>
>>
>> Romascanu, Dan (Dan) wrote:
>>>
>>>> -----Original Message-----
>>>> From: [e-mail] [mailto:[e-mail]] On
>>>> Behalf Of Andy Bierman
>>>>
>>>> We have also been using XSD for the canonical XML syntax
>>>> representation in protocol documents for quite some time.
>>>> I strong object to changing this policy to RelaxNG or DSDL
>>>> without IETF Approval first.
>>>>
>>>>
>>> Allow me to express shortly my view as a contributor.
>>>
>>> I do not believe that we deal here with a change in policy. The
>>> canonical XML syntax representation stays with us as the 'on the wire'
>>> interoperability level. We are introducing an operator-friendly language
>>> that is described by Randy in his preliminary charter draft and the
>>> mapping between the operator-friendly level and the XML. These would be
>>> the mandatory-to-implement elements of the framework
>>>
>>> We also recognize that there is a place for a machine-friendly
>>> representation that would accommodate existing tools and allow for
>>> import of existing XML schemes already defined by other WGs in the IETF
>>> or other SDOs, in RelaxNG or DSDL or other. Mappings between the
>>> operator-friendly language and RelaxNG or DSDL would be optional, but at
>>> least one of them will be a loss-less translation.
>>>
>>
>> This sounds fine from a policy POV.
>> In practical terms, it seems that document reviewers
>> will have a lot more work to do, verifying that:
>>
>>    - all these mappings are well-formed
>>    - they match the semantics of the high-level model
>>    - there are no conflicts between any of the machine-mappings
>>    - all DM variants in the document describe the exact same set of XML
>>
>> Since the IETF track record for validating XSD correctness
>> is so poor already, how will the box score look when 3 complicated
>> DMLs need to be reviewed, instead of 1?
>>
>> Or is the intent to fill up NETCONF data model drafts will lots
>> of pages that nobody will review?
>>
>>
>>
>> Andy
>>
>> _______________________________________________
>> NGO mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/ngo
>>
> 
> _______________________________________________
> NGO mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/ngo

-- 
Balazs Lengyel                       Ericsson Hungary Ltd.
TSP System Manager
ECN: 831 7320                        Fax: [phone]
Tel: [phone]     email: [e-mail]
_______________________________________________
NGO mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/ngo