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

Ladislav Lhotka wrote:
> tom.petch píše v St 12. 09. 2007 v 10:17 +0200:
> 
>> I agree with you that XML/XSD is the only realistic alternative.  I have
>> immersed myself in it over the life of NETCONF and the more I learn the less I
>> like; the fundamentals are plain wrong (IMnotsoHO).
>>
>> But, like other not quite perfect pieces of technology in the past, it has swept
>> most before it so I think it the only choice.
> 
> I really don't understand this. Virtually everyone is complaining about
> XSD being broken and I've never heard similar objections being raised
> against RELAX NG, but apparently nobody seems to give it a slightest
> consideration(?). As a data model specification language, RELAX NG is
> IMO much more readable than XSD and I wrote a XSLT stylesheet allowing
> annotated RELAX NG schemas to be incorporated in xml2rfc sources in a
> way that resembles Knuth's literate programming. So what's wrong with
> it?

I don't think that many people know RelaxNG.
When <appinfo> and <documentation> type of clauses are included,
it isn't so clean anymore.  It may be just as hard to learn for
people as XSD.  It doesn't really deal with any NETCONF-specific
issues (like max-access and conceptual tables) any more than XSD.

Picking a DML is very subjective.  I doubt there are that many
objective metrics you could apply.  Metrics like quantity of
deployed tools can only go so far.  Human usability is very hard
to quantify.  Varying additional factors, like SMIv2 reuse,
CLI integration, application complexity, and internal SW-dev
environment make objective evaluation even harder.

The only important factor (applying Occam's razor here) is that
WG members who are trying to standardize a NETCONF data model have a
common understanding of the entire document, which is high enough
to maintain a reasonable standards process.   Everything else is secondary
to that goal.  This is a standards body.  We make standards.
That is our product.  If we cannot understand the documents we
are trying to standardize, then we have no product.

I do not really care that much which DML is chosen for NETCONF,
as long as this primary goal can be achieved.  It is clearly
not be achieved with XSD.  That is why (IMO) the long term
solution is a new DML that can be translated to XSD without
information loss (<appinfo> not considered info loss).

> 
> Of course, specific NETCONF software could use XSD or whatever schemas
> automatically translated from RELAX NG, if necessary, so this discussion
> is really about data model specifications intended to be read by humans.
> 
> Lada
>  

Andy

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