Frederic,
On 2005 Jun 9 , at 13.30, Frederic V. "Rick" Hessman wrote:
> http://monet.uni-sw.gwdg.de/cgi-bin/xml_validation/scripts/
> createUCDSchema.pl
Cough.
Unless I'm mistaken, your pattern there admits UCDs like "pos.eq.ra;instr.beam;meta.ucd". If we follow the logic of your approach, which wants Schemas to do all the validity checking, and so exclude such garbage UCDs, then you'll have produce an XSchema fragment which lists all the complete UCDs anyone might legitimately ever want, which I hope you would agree was unlikely to be useful. If you think that is going too far, then you agree that there will have to be some semantic validation of UCDs _after_ syntactical analysis, in which case there's no need to do any syntactical analysis beyond confirming that the UCDs conform to the production in the UCD standard.
I'm afraid I don't understand what problem you're addressing. You appear to be thinking of some situation where there will be syntactical analysis of a UCD in a FITS header or an XML attribute, but no subsequent semantic analysis or use of that UCD, which is where garbage UCDs would naturally be detected and usefully reported.
> I sense that there is some reluctance to have the UCDs parsed, though
There's no reluctance on my part to have UCDs parsed, but they're parsed into character sequences, dots and semicolons, for separate subsequent semantic analysis (what does this UCD mean? does it mean anything?). When I parse the famous `colourless green ideas sleep furiously' I can do so perfectly successfully; attaching meaning to it is not my parser's problem.
> I haven't yet heard any good reason for allowing garbage UCDs in
> documents.
This isn't an argument for allowing garbage UCDs anywhere. It's simply the argument that XML syntax-analysis tools have nothing to do with the semantic validation of UCDs: just because XSchema is a hammer, it doesn't follow that everything else is a nail.
The premature semantic validation you're talking about would not improve data security; would make the system less, not more, robust; and be on the whole less usable, by reporting semantic errors at an inappropriately low (ie, syntactic) level.
UCDs are not XML.
All the best,
Norman
-- ---------------------------------------------------------------------- Norman Gray : Physics & Astronomy, Glasgow University, UK http://www.astro.gla.ac.uk/users/norman/ : www.starlink.ac.ukReceived on 2005-06-10Z15:48:38