[ivoa-std/SSA] Invalid UCDs : proposal for an erratum to SSA 1.1: instr.fov

[Markus Demleitner]

Dear Mireille,

Thanks for tackling this.  I by and large like the proposed
resolutions of the spec conflicts, somewhat reformatted here for the
benefit of MUAs preferring text/plain:

On Mon, Oct 04, 2021 at 05:18:46PM +0200, Mireille LOUYS wrote:
> Char.SpatialAxis.Coverage.Bounds.Extent
> phys.angSize; instr.fov
> Aperture angular size
> double
> 
> Char.SpatialAxis.Coverage.Support.Area
> pos.outline;instr.fov
> Aperture region
> char *
> 
> Char.SpatialAxis.Coverage.Support.Extent
> phys.angArea; instr.fov
> Field of view area, sq. deg.
> double
> 
> 	
> Rationale :
> /instr.fov/ represents a geometrical shape on the sky, where an instrument
> is observing  a signal.

Not that it's dramatically important, but I don't think we need
strong statements like this that conflict with the later statement
(which I agree with) that instr.fov should be read as "related to the
field of view of an observation" or, if we want to avoid recursive
definitions, perhaps "related to the part of the object observed
covered by an observation" (ugh).

Anyway, I'd just say:

  UCD requires instr.fov to be secondary.  Hence, the UCDs for the
  fields with utypes ... as prescribed by SSAP 1.1 are invalid and
  need to be repaired.

For me, that's quite enough to motivate why we're touching the thing.
Based on that, I think all that's required in addition is a single
sentence:

  This erratum provides valid UCDs, which not only have proper
  semantics but also enable UCD-aware clients to determine the nature
  of the fields if they do not know the SSAP data model.

Done, for all I care.

Well, except that an erratum should also come with an impact
assessement.  For that, I'd propose:

  SSAP clients do not identify fields by UCDs and hence are
  unaffected by this change.  Clients relying on UCDs to determine
  how to treat individual columns will now be able to gain some basic
  understanding of the affected columns, which they could not before
  (as the UCDs were invalid and hence meaningless).

  Most importantly, valid SSAP services will raise fewer errors in
  UCD validation.

> UCDList EN-1.4 defines /instr.fov/ as
> S | instr.fov | Field of view
> 
> This definition is fuzzy, and can be clarified and replaced by
> S | instr.fov | Related to Field of view
> for the next UCDList update UCDList1.5.
> 
> Proposed Action : change definition of instr.fov to Related to
> field ov view .

I'd say we should just apply that change in a PR to UCDList.  Let's
not burden the erratum with that, as errata to two standards at the
same time are... complicated (to avoid saying "we've not planned for
them"), and I don't think there's a particular urgency to get in that
"related";  UCDList 1.5 should also be out relatively soon -- I'll
personally poke the TCG folks this time around.

Thanks again,

           Markus