TAPRegExt revision

[Mark Taylor]

Markus,

your changes (a) and (b) sound like they could be reasonably be cast
as "errata" rather than actual (even minor) changes to the standard.

If that's the case, then the document and schema version number need
not change.

The TCG started to agree on a relatively lightweight formal process
for errata at the F2F meeting in Madrid.  I'm not sure that all the
details of that process have been worked out (IIRC a new DocStd
document is under preparation), but this alteration to TAPRegExt
would seem like a good candidate for an early client of it.

Mark

On Fri, 30 May 2014, Markus Demleitner wrote:

> In February (thread starting at
> http://www.ivoa.net/pipermail/registry/2014-February/004909.html), we
> briefly discussed the impact of having "good" standardIds on
> TAPRegExt; in TAPRegExt's schema, the data model is a
> vr:IdentifierURI, which doesn't fit the new, post-StandardsRegExt
> standard identifiers.
> 
> Back in February, I think the conclusion was that TAPRegExt should
> simply have xs:anyURI here.  We'd have to go ahead with this now, as
> RegTAP would already use this.
> 
> So, I'd prepare a new version of TAPRegExt with the following two
> changes:
> 
> (a) tr:dataModel/@ivo-id would be of type xs:anyURI rather than
> vr:IndentifierURI
> 
> (b) in the example, the (currently bad) standard id of ObsCore will
> be fixed.
>
> Now -- if anyone has other things to fix with TAPRegExt, this would
> be a good time to speak up.
>
> In connection with (a), I have another flamboyant itch: The namespace
> of the schema, http://www.ivoa.net/xml/TAPRegExt/v1.0, is probably
> built into several clients out there, as the need to interpret what's
> on the TAP servers' capability endpoints.  Changing that namespace
> URI is therefore expensive.  Given that clients in all likelihood
> will interpret the data model URIs as basically opaque strings: Would
> anyone seriously object to updating TAPRegExt without changing the
> namespace URI?
> 
> There is one catch with not changing the URI: If we change the
> "standard" schema file name (to, say, TAPRegExt-v1.1.xsd) but not the
> tns URI, things would be confusing.  I guess the fix could be to not
> change the file name or to just have a comment next to the tns
> declaration.
> 
> Cheers,
> 
>          Markus
> 
> 

--
Mark Taylor   Astronomical Programmer   Physics, Bristol University, UK
m.b.taylor at bris.ac.uk +44-117-9288776  http://www.star.bris.ac.uk/~mbt/