SimpleDALRegExt 1.1

[Markus Demleitner]

Dear DAL, dear Registry,

  ---------------------------------------------------------------
  TL;DR: SIAv2 should have a Registry metadata schema.
  http://docs.g-vo.org/SimpleDALRegExt.pdf allows to just adopt
  SIAv1's.  If you think we need more, speak up now
  ---------------------------------------------------------------

Currently, there is no way to furnish the registry records of SIAv2
services with additional metadata -- this is particularly regrettable
because this way, automatic validators have no reliable test
query[1].

In the late stages of SIAv2 standardisation, there was a tentative
agreement (I think) that the metadata model that SimpleDALRegExt made
for SIAv1 -- essentially

 	validationLevel*, description?, interface*, imageServiceType,
 	maxQueryRegionSize?, maxImageExtent?, maxImageSize?, maxFileSize?,
 	maxRecords?, testQuery?

-- will in principle work for SIAv2.

NOw, SimpleDALRegExt 1.0 fixes the standardIDs for the capability
types it defines.  That's a technicality that make it so that if you
write in your registry record:

  <capability xsi:type="sia:SimpleImageAccess"

you currently have to go on with

    standardID="ivo://ivoa.net/std/SIA">

whereas for SIAv2 we would like to say

    standardID="ivo://ivoa.net/std/SIA#query-2.0">

To fix this, I've prepared SimpleDALRegExt 1.1 [2] in volute,
http://volute.g-vo.org/svn/trunk/projects/registry/SimpleDALRegExt
(which is also where you'll find the fixed schema).

A formatted version is at http://docs.g-vo.org/SimpleDALRegExt.pdf.

The essential change is that the standardIDs for the legacy DAL
services are now defined in the text, whereas the schema no longer
fixes (or indeed gives) them.  There are a few editoral changes as
a consequence.

Two minor updates also went in: 

(1) I'm now declaring the schema files in the repo as authoritative
and no longer include the full schema.  I believe that's a fairly
necessary consequence of Paul's schema versioning policy proposal
http://ivoa.net/documents/Notes/XMLVers/ (which I fully intend to
exploit here by not changing any namespace URIs).  Also, I think
printing schema files is an insult to trees and whatever was turned
into the soot your toner  is made of.

(2) I updated the example and a recommendation for the discovery of
standard services to the system based on capability/@standardID that
has evolved over the years (rather than using interface/@version).


So:

* Are there major objections to what I'm doing here?

* Does anyone want additional SIAv2 metadata?  Note that this
  standard also determines what SIAv2 services can declare on their
  /capability endpoints.

* Does anyone have updates to the SCS, SSAP, or SLAP metadata
  schemes?

* Are there any operators of SIAv2 services who would try this?

As always you're most welcome to fix problems in volute when you
spot them.

I'd suggest that any followup discussions should take place on
registry.

Cheers,

          Markus


[1] I would assume that MAXREC=1 would be a suitable test query for
all SIAv2 services, though.  Is that right?  Either way, test queries
aren't the only reason to have the registry extension.

[2] You will have noticed the commit messages if you've subscribed to
http://lists.g-vo.org/cgi-bin/mailman/listinfo/volutecommits