Timesys note review

Mark Taylor M.B.Taylor at bristol.ac.uk
Thu Dec 6 11:29:04 CET 2018

On Thu, 6 Dec 2018, François Bonnarel wrote:

> Le 06/12/2018 à 10:50, Mark Taylor a écrit :
> > On Thu, 6 Dec 2018, François Bonnarel wrote:
> > 
> > > The main issue there is not (to my eyes) that ref can come in several
> > > contexts
> > > but that there is one single ref allowed on each element, which is coming
> > > from
> > > xml itself and not from VOTable properly.
> > No, it is a VOTable constraint.  VOTable could have been written so
> > that you could make multiple references from a single element,
> > e.g. by giving the ref attribute the type xs:IDREFS rather than
> > xs:IDREF.
> Thank you. I didn't know that, and forgot about xs:IDREFS
> > But it's not.  In principle it would be possible to
> > change that without invalidating any existing VOTable instances,
> > since anything which fits IDREF also fits IDREFS.  But I'm not
> > proposing that we do that.
> Can you say why ? It looks like  that could avoid using too much GROUPs in
> some cases. But maybe there are other drawbacks I don't see.

Mostly I was just being conservative; changing standards is usually
somewhat painful.  Although it wouldn't invalidate existing table
instances, if VOTables started to have multiple tokens as values
of the ref attribute, it could confuse existing software that's
not expecting to see that.  But, if we thought it was going to
solve significant problems we could consider making this change
for VOTable 1.4.  Off the top of my head I don't see any terrible
problems it would cause.  However, as somebody else on this thread
has also mentioned, the semantics of ref/ID use in VOTable is not
well defined or documented, and allowing multiple references would
tend to make that worse.  So if we did make this change (and maybe
even if not) I would like to see some better documentation,
probably in the VOTable document, of what such ref/ID relationships
actually mean when they are used between various pairs of elements.

Mark Taylor   Astronomical Programmer   Physics, Bristol University, UK
m.b.taylor at bris.ac.uk +44-117-9288776  http://www.star.bris.ac.uk/~mbt/

More information about the voevent mailing list