David and Mark,
On Mon, 26 Jan 2004, Mark Taylor wrote:
> On Mon, 26 Jan 2004, David Berry wrote:
>
> > Mark,
> > Thanks - that's probably saved us a good number of being-hours! I am
> > anxiously awaiting developments on the off-shoot "Quantity tables" thread
> > of Linde/Shaya. The area of overlap between VOTable and data models is a
Ditto. It means I can delete all those VOTable messages I haven't yet
had the heart to plough through. Phew!
> > Having achieved a certain level of agreement in the Thomas/McDowell/Berry
> > debate on Quantity, Brian now wants to go on immediately to decide on
> > serialisations. It seems to me that many people (such as you and Al) are
> > expecting that the data model should be expressable as a VOtable of some
> > flavour. Seems reasonable. However, Brian's initial suggestions which he
>
> I'm not up to date on the data model stuff, but I wouldn't say I was
> expecting a Quanity to be expressible as a VOTable. Might be sensible,
> might not, but I certainly wouldn't start from the assumption that it is.
> To me, a VOTable is a good way of expressing tabular data, and I
> didn't think that a Quantity was going to be particularly tabular.
> If you're coming up with an XML-based serialization format it
> might be a good idea to specify it with reference to VOTable so that
> similar things look similar though.
I also wouldn't expect a Quantity serialisation to necessarily
resemble VOTable. In fact, I think I'd go further and expect it to be
rather orthogonal.
What I would expect it to look like is AST's object model with angle
brackets round it. How about
<object>
<mapping invert="0" nin="2" nout="2">
<slamap cvt="amp" args="2004.5 2000.0"/>
</mapping>
</object>
You and Rodney have already produced a serialisation of WCS (which I
know is only part of the Quantities fight); you only have to reexpress
it as XML. If you say `look, we've already _done_ this!', then
Jonathan and Brian would have that to respond to.
The problem with XML is that folk let loose with XML Schema editors
seem always to go completely bananas and produce huge wobbling schemas,
seduced by the pretty boxes on the screen and the impressive quantity
of schema text which the tool emits. With XML, less is very much more, and
designing XML is much more a matter of good taste than specific knowledge
(there really isn't an awful lot to know).
My rule of thumb is that if anyone suggests XML Schemas as opposed to
DTDs or RELAX-NG, then either they're concerned with sucking XML into
databases (or some equally sophisticated reason), or they know veeery
little about XML. Harsh? I think so, yes.
See you,
Norman
--
---------------------------------------------------------------------------
Norman Gray http://www.astro.gla.ac.uk/users/norman/
Physics and Astronomy, University of Glasgow, UK [log in to unmask]
|