8/5/02.
Ed, Kirk, Roy, et al,
Ed Shaya wrote:
> I don't know if you looked at XDF before you wrote this e-mail,
No, I've not yet looked at the XDF documentation (an omission which I'll now rectify! - it sounds very relevant). The things that I had in mind were the VOTable itself and the (non-XML) hierarchical format for bulk data which Starlink has used for many years.
> but
> you're strawman array and parameter elements are right on to the XDF way
> of doing things. We can discuss some details that we would do slightly
> differently (like put each dimension size on the relevant axis element),
> but I think we are in fairly good agreement on the basic approach and
> the need to start working out a VO standard. Perhaps we need a working
> group specifically for this.
The XML elements that I suggested, VOBulkData and VOHDF, were just drawn up quickly off the top my head as illustrations. I wouldn't particularly defend details, such as where I put the specification of the dimension size of each axis. Though we should (and doubtless will!) discuss such issues.
Roy Williams wrote:
> ... I see a collection of XML document
> definitions (aka classes) that can be extended through inheritance -- for
> example a VOTable that has XDF objects in the table cells. I see development
> next of Image and Spectrum documents (among others) which inherit from an
> N-dimensional Datacube object.
I'd implicitly assumed that there would be separate documents defining the VOTable and (in my notation) the VOBulkData and VOHDF elements, so that work on the specifications can proceed asynchronously and they can be used separately, as required. This approach sounds reasonably similar to what Roy has in mind.
regards,
Clive.
Clive Davenhall Institute for Astronomy, e-mail (internet, JANET): acd @ roe.ac.uk Royal Observatory Edinburgh, fax from within the UK: 0131-668-8416 Blackford Hill, Edinburgh, fax from overseas: +44-131-668-8416 EH9 3HJ, Scotland.Received on 2002-05-13Z07:01:25