More UCD woes

Markus Demleitner msdemlei at ari.uni-heidelberg.de
Wed May 4 16:43:43 CEST 2022


Dear DAL WG,

I'm sure you've all been waiting for my next nitpicking of UCD use in
DAL.  Wait no longer.  Except: this time, the cut is a bit deeper.

SIAP 2 in section 3.1 ("Successful Query") says:

  The response from a successful call to the {query} resource is a
  table consistent with  ObsTAP responses as described in [7].
  The ObsCore data model specifies all the (VOTable) field names,
  utypes, UCDs, and units to use in the response, as well as which
  fields must have values and which are allowed to be empty (null).
  The {query} response must contain the required ObsCore fields and
  may contain additional fields (from [7] or custom fields from the
  service provider). Examples 

[7] is a reference to Obscore 1.0, version-sharp.

Which is where the trouble starts.  Because, between Obscore 1.0 and
Obscore 1.1, the UCD of the s_region column was changed from
phys.area;obs to pos.outline;obs.field.

Ever since I've moved by Obscore schema to 1.1, my SIAP 2 response
has thus be invalid (brown bag for having ignored that for so long;
cheers to Renaud for making it straightforward to have a simple
command for "how do my services do"?).

I *could* rather easily repair that by fiddling in the old UCD if
DaCHS realises it is talking SIAPv2, but: Is this what we want?
Shouldn't SIAPv2 just say "any response compliant to Obscore version
1 is fine"?  And if not, why?  Would we then issue a new version of
SIAPv2 whenever there's a new version of Obscore?

        -- Markus


More information about the dal mailing list