VODax and VODMX

[Jonathan McDowell]

> Tony Linde wrote:

> In order to curtail the rancorous debates on VOTable, I've taken Roy's
> suggestion and have set up an effort to develop a data model-based
> alternative to VOTable. I've tentatively titled it the VODaX (Virtual
> Observatory Data eXchange) format and the start page on the wiki for it is:
>   http://www.ivoa.net/twiki/bin/view/IVOA/VODaX

Tony, I guess I'm a little puzzled by this since I think we are already
doing it.  I would have thought that the content of VODaX is "Easy, we'll use
schema-based XML to exchange data" and the content of VODMX is addressed
initially by the serialization sections of the Quantity data model
document, and in general by the discussions on schema serializations
that have been going on for some time in the DM group.  We have proposed
XML serializations for Quantity, for Space-Time Coords, and (at a
tentative level) for Mapping, while a proposal for Spectral Energy
Distributions is circulating within the DAL group SSAP effort and will
be unveiled in the next couple of days. We certainly need help
developing these and other serializations, and I would welcome your
comments on any of the existing proposals.

However, I am not convinced that the serialization effort should be separated
from the DM effort in the way that you appear to suggest, because it is often
hard to understand the DM abstractions without serialization examples,
so the design loop is pretty tight.

My attitude to the serialization problem has been that

(1)  I expect to define a reference serialization for each data model in
XML schema, as has been stated by the DM group in each of its WG
meetings.

(2) I also expect VOTABLE serializations to be used, at least for some
data models, and the UTYPE effort is trying to develop a standard
way to do this.

(3) In at least some cases we may even define a reference FITS serialization if
that seems appropriate.

All of these are part of the existing DM effort. Perhaps there is some
content to your VODaX idea that I'm not understanding - feel free to
clarify. And I'm open to having a separate twiki area for the XML schemas
derived from the data models, we can call it VODMX if you like, although
we should make the links to the corresponding models very clear.

In parallel with this is the contentious argument of when to use VOTABLE
and when to use object-specific schemas. I think this is an argument we
need to have, and the DM group is certainly one legitimate place to have
it. Ultimately I believe the argument will be resolved only after
implementations - the two camps both have legitimate points of view
that reflect a fundamental division in the VO between 'evolve current
astronomy software and approaches so we can interoperate' and
'use the latest software technology even if it doesn't map well
to how astronomers currently do things'. 
As you imply, the solution for now is to let both camps develop
their standards. Yells of 'heretic!' and other suggestions that
one or other approach is fundamentally flawed will continue to
fall on deaf ears and will be unproductive; better to work with people
who agree on your approach and develop that approach, than to
spend your time derailing others' efforts.

I don't know yet which approach will turn out to be most practical,
so I'm happy to work to push along both paths.

  - Jonathan