XSD - new VALID schema

[Tony Linde]

> If there are problems with the specification, one shouldn't 
> have to call them out by referring to a schema.  And, in 

I'd agree with that the schema should reflect the spec.

> fact, there is no one single schema that is acceptable.  One 
> shouldn't say "the VOEvent schema", but rather, "a VOEvent schema".

I don't agree with that. The VOEvent team should produce a schema which
completely reflects the spec, the schema should be part of the spec document
and that is the definitive schema for VOEvent. AFAICR that is the way W3C
writes its specs/schemas.

> versions of VOEvent, post-1.2, but would strongly urge we not 
> modify v1.1N for "technology" reasons.  I think the RFC 
> comment period should be used to call out missing or broken 
> features, not to focus on issues of software philosophy.

The RFC is for people to raise ANY issues which they think is a problem in
the spec whether it is functional or technical or anything-else-al. And the
workgroup then have to reply to those issues.

I would say though that such issues should all be listed on the RFC wiki
page with the responses so that when the TCC and exec look to see if the
spec is really ready to be passed on to the next stage, they can judge if
all issues have been properly addressed. It seems that most stuff has been
posted there alongside these discussions so far but it'd help if people
posted to the wiki page first and then drew attention to it here with the
workgroup adding the response after discussion has concluded.

Can I ask those who have raised issues (hint, hint, Petr) to make sure they
are all on the wiki page.

Cheers,
Tony. 

> -----Original Message-----
> From: owner-voevent at eso.org [mailto:owner-voevent at eso.org] On 
> Behalf Of Rob Seaman
> Sent: 10 August 2006 17:07
> To: Petr Kubanek
> Cc: voevent at ivoa.net
> Subject: Re: XSD - new VALID schema
> 
> Hi Petr,
> 
> If I understand, your previous message was offering a new 
> schema that correctly captures the specification, but this 
> latest message is suggesting changes to the specification to 
> agree better with XML schema best practices (or schema 
> limitations, perhaps)?
> 
> I think we should be willing to entertain such for future 
> versions of VOEvent, post-1.2, but would strongly urge we not 
> modify v1.1N for "technology" reasons.  I think the RFC 
> comment period should be used to call out missing or broken 
> features, not to focus on issues of software philosophy.
> 
> As far as the question of schemata in general,  a schema that 
> attempts to capture any specification should be tight fitting 
> where the specification is stringent and loose fitting where 
> the specification is accommodating.  A schema should not be 
> more restrictive than the specification - which is to say 
> that all conforming documents should validate.
> 
> If there are problems with the specification, one shouldn't 
> have to call them out by referring to a schema.  And, in 
> fact, there is no one single schema that is acceptable.  One 
> shouldn't say "the VOEvent schema", but rather, "a VOEvent schema".
> 
> Rob
> 
>