Proposed changes to VOTable

[Mark Taylor]

Dear IVOA members,

As decided by the TCG, a special session was scheduled at the Pune
Interop, coordinated within the Applications WG, to discuss recently
raised issues with VOTable.  These arose from concerns voiced
initially by Tom McGlynn in relation to use of VOTable for TAP
result output, and summarised on the VOTableIssues wiki page.
The presentations are visible near the bottom of the 
InterOpOct2011Applications wiki page.

Following discussion by those present, we came up with some
recommendations for a VOTable 1.3.  The problems being addressed
and proposed solutions are summarised in final three slides of
the Apps WG closing plenary presentation, at

   http://www.ivoa.net/internal/IVOA/PlenarySessionsOctober2011/apps-review.pdf

To summarise the summary:

  + There should be a VOTable 1.3, incorporating the following changes:

      - For the TABLEDATA encoding:

         * Empty TD element permitted in integer columns to represent NULL
           (previously was illegal)

         * Empty TD element in floating point columns to represent NULL
           (previously meant NaN)

      - A new BINARY2 encoding:

         * Like BINARY, but with an additional per-row bitmask header
           indicating whether each cell is NULL or not.

      - A new optional "serialization" parameter to the VOTable MIME type
        to distinguish between the (now) 4 encodings: TABLEDATA, BINARY,
        BINARY2, FITS.  (Parameters, optional or not, are a standard
        MIME type feature - see RFC2046).

  + Since the BINARY2 encoding will not be comprehensible to existing
    VOTable parsers, VO services will not deliver it except to clients
    which explicitly request it.  The mechanism for such a request
    will be service-dependent, but will likely make use of the newly
    parameterised MIME type.

Opinions on these recommendations are solicited now from anyone not 
present at the Pune Interop.  Given the history of this issue,
in particular the opinions of Tom McGlynn, who raised this in the
first place, and Francois Ochsenbein, as last chair of the currently
dormant VOTable WG, are welcome.

This message was posted on the interop mailing list in view of the
wide effect that a change to the VOTable standard would have across
the IVOA standards.  Followups however should be made to the
votable at ivoa.net mailing list.  If/when conclusions are reached,
as summary will be posted to the interop list.

If consensus is achieved the TCG will consider how best to move 
towards a VOTable 1.3.

Mark Taylor (Applications WG Chair)

--
Mark Taylor   Astronomical Programmer   Physics, Bristol University, UK
m.b.taylor at bris.ac.uk +44-117-928-8776 http://www.star.bris.ac.uk/~mbt/