Please review: VOResource PR #2

[Markus Demleitner]

Dear Registry folks,

Between the Interop and other commitments, I had totally forgotten
our planned fast-track VOResource update to accomodate the UAT
(context:
<http://mail.ivoa.net/pipermail/registry/2022-October/005504.html>).

In a way, that was fortunate, because that way I remembered in time
there were two items on VOResource-Next:
<https://wiki.ivoa.net/twiki/bin/view/IVOA/VOResource-1_1-Next>.

The first one, forcing data providers to have proper HTTP URIs in
referenceURL, is what PR#2,
<https://github.com/ivoa-std/VOResource/pull/2>, addresses.

Technically, this would be a breaking change, because records
formerly valid would become invalid.  I would, however, argue that
these records were be factually broken already because a client will
not be able to open whatever the data providers put in (without wild
guesses).

Also, no actual record in the Registry would be affected at this
point (ok: this is because I have asked data providers to fix their
reference URIs in spring, but that no new bad referenceURLs went in
since then suggests that this is a reasonable requirement to have).
Try it yourself:

  select reference_url from rr.resource 
  where 
    not reference_url is not NULL
    and reference_url like 'http://' 
    and not reference_url like 'https://'

on <http://dc.g-vo.org/tap>.

I note in passing that there are currently five resources with
NULL/empty reference URLs.  But these are already invalid; if you
publish one of them (authorities mssl.ucl.ac.uk, svo.ifca, vo.sim),
please fix them.

So... please comment on PR#2 (or protest here); if I don't hear from
you, I'll force-merge about a week from now.

This would leave the second item of VOResource-Next, defining a
keyword for DOI sources.  I plan to allow "doi" in format, which
would then imply a content with a plain DOI (i.e., neither the http
URL form nor the URI form from altIdentifier).  If you have thoughts
on that, please share them.

Thanks,

         Markus