Jonathan, Patricio et al:
Regarding Jonathan's example of a table with measurements of flux density
at 3 (or more) frequencies, and Patricio's comments,
http://www.ivoa.net/forum/ucd/0006.htm
I indeed think we should get away from PHOT_FLUX_RADIO_1.4G or anything
like that, and instead have some way of identifying a flux density
measurement with the frequency at which it was made, without having a
separate UCD for every frequency anyone uses. In the long run
Ray's proposal http://www.ivoa.net/forum/ucd/0011.htm to use Xpath seems
exactly right, that is what I was trying to say in words (i.e. we need
values and attributes etc.).
However, is there a means of transition from the present use of UCDs, e.g.
by associating columns
FREQ FLUX_DENSITY FLUX_DENSITY_ERROR
GHz mJy mJy 1.4 5 1 4.8 10 1 22.0 20 3
or whatever?
I suppose there are two problems here; firstly, to respond to Sebastien's summary of the present discussion around UCDs - to see how much consensus there is on what one would like in an idea world - and there seems to be general agreement on the most general data models at least, different groups may need more detail in different aspects but all the discussions seem to be moving in the same direction.
Secondly, however, how much can be done 'seamlessly' based on the existing, working CDS UCD system? It is important to keep that going, not just because astronoemrs depend on it and want to see it do these things they hear about in reports from VO meetings, but because we need to test developments on the community as much as possible, and we will find subjects much more easily if we can give people prototypes which look as familiar as possible (as with the AVO demo). Also many new functions e.g. the SED tool and VOPlot are being developed based on the existing UCDs. Is this possible, is this the direction in which IDHA is going? Or will there come a break point, and if so, when?
cheers
a