Cube Data Access Layer implementation note

[Markus Demleitner]

Dear DAL,

On Thu, Mar 03, 2016 at 10:06:09AM -0800, Patrick Dowler wrote:
> Is adding a separate endpoint (new capability, new standardID) better?
> 
> This sounds somewhat like the metadata capability that we postponed until
> after the dot-0 versions of the minimal specs. It seems to me like an
> increase in scope and there is too much uncertainty here... so IMO we
> should be OK with TAP|SIA -> DataLink -> SODA or the optimised TAP|SIA ->
> SODA. If a provider can't do the latter, it is only because you already
> need a DataLink service.

FWIW, I heartily agree with this -- none of the three scenarios (see
http://mail.ivoa.net/pipermail/dal/2016-January/007281.html) that we
seem to have converged upon at least in content if not in form (cf
http://mail.ivoa.net/pipermail/dal/2016-February/007325.html) require
"SODA self-description", and I maintain there are fairly strong
arguments against needlessly clobbering the ID-only response from a
SODA endpoint and defining another IVOA service type to be defined
and validated.

So, I propose to say

  "The response from a SODA endpoint to requests giving no parameter
  at all or just an ID is undefined in this version of the
  specification.  Both cases are reserved for standards evolution."

at an appropriate point in the spec.

Cheers,

       Markus