On Thu, 8 May 2003, Norman Gray wrote:
> I've prepared an HDX overview, which is currently sitting at
> <http://www.astro.gla.ac.uk/users/norman/star/java/hdx/>
>
> That is the line I plan to take on HDX in the IVOA talk, which I
> intend to write tomorrow. So:
>
> * Do you think this is the right line? I've tried to aim it at what I
> believe this audience to be (and you'll surely note that I haven't
> said `ontology' once, nor even `semantics', though I have at one
> point said `architecture'), but I'm not honestly confident that
> I've succeeded.
>
> * I plan to at least read over it again tomorrow, but don't have
> current plans to rewrite or much extend it. So if there are any
> egregiously opaque bits, make sure I haven't glazed over them.
Hi Norman,
thanks, that looks like a good summary of what HDX is. As for whether it's
the right line, who knows, depends on who's listening.
If you were to extend it for the less-technical I'd say it's missing an
up-front statement about what problems HDX is solving and why it's
implementation is more elegant and suitable for the web than traditional
data forms.
BTW, I did see a "provenance" slip in there. That particular paragraph is
a little opaque. My guess is that this is a passing reference to the
obviously missing part of HDX, a real place to put miscellaneous meta-data
(c.f. FITS headers)?
> * Unless folk shout otherwise, I plan to mention this overview
> on the [log in to unmask] mailing list tomorrow. I am relived that Tim
> also thinks this is a good idea.
There was an official appeal for people to do this.
> * It's not _completely_ clear to me how I demo this. I suppose I
> could give the SoG-plus-photom demo (if I can learn how to do it on
> Monday), and say, `this is passing HDX wrapped in SOAP, so look,
> we really _have_ written this code'. But is that really going to
> blow anyone away? The Treeview demo does show the structure, and
> is generally regarded as a Good Thing, so perhaps that would be saner.
I'd vote for the Treeview demo. It's clearer and if you can get to hook up
to SoG and SPLAT, that's passing HDX information between applications
using web service-like mechanisms (endpoints, plus SOAP).
(p.s. in the examples there's a line:
<x:image uri="file:/tmp/mydata.fits"/>
remind me that this is OK and the uri attribute isn't lost during some
namespace transformation and should really be x:uri).
Cheers,
Peter.
|