WD-DALI-1.1

[Markus Demleitner]

Hi Pat,

On Fri, Sep 09, 2016 at 08:35:56AM -0700, Patrick Dowler wrote:
> Those are my two motivations for defining xtype='region" in DALI-1.1
> (which we need for SODA-1.0 anyway due to other xtypes introduced):
> consistent use.

...though, of course, we've identified that we're fairly unhappy with
the way TAP 1.0 serialised geometries in VOTable (and therefore
change it), and I felt that my uneasiness with POS has spread a bit
further.  If these impressions are right, I'd say let's not put stuff
into a new document that at least a sizable number of people want to
phase out. Unless there's a strong use case, of course.

Consistent use doesn't appear to fall under "strong use case" to me;
I think I'd rather call it standards cosmetic.  While self-consistent
standards are certainly desirable, leanness (in the sense of:
as little as possible specified that's not strictly required for
interoperability) is more important to me.

There's a practical problem, too: There's quite a few services
deployed that use (at least) adql:REGION already for something
incompatible with what DALI says (STC-S with simple metadata as per
the non-normative TAP 1.0 appendix).  Even though that use of the
adql:REGION xtype was done under some sort of experimentation clause,
these things are going to linger around for some time to come, and
I'd be reluctant to (semi-)overload xtype="region".

If we find out we cannot live without xtype="region" in DALI 1.1, we
need to say at least something like: adql:REGION is the experimental
one, region without prefix is the DALI one (or so).

          -- Markus