Re: Sessions at Trieste

From: Patrick Dowler <patrick.dowler-at-nrc-cnrc.gc.ca>
Date: Mon, 12 May 2008 10:52:43 -0700

On 2008-5-12 07:13, Pedro Osuna wrote:
> Dear all,
>
> there is one single session for VOQL to present the ADL document.
>
> I would like to more or less repeat what we did in Cambridge last
> time, and have:
>
> - Inaki presenting the overall ADQL
> - Jeff presenting the BNF and parser
> - Pat presenting the Region
>
> Would you guys agree on that?

Since we are in the RFC, it seems like we are inviting discussion of the whole spec. I feel it would be better to have a session where we air the specific comments made on the RFC page and answer them as well as we can.

> It would also be nice to hear from you on a previous mail I sent
> regarding the collection of inputs for the RFC from the different
> communities, as that should also be reported in Trieste.

The comments so far are from bob (as NVO?), Roy (as TCG chair?), and Francois (who is already on the TEG).

Aside from some lesser issues, the comments from both Bob and Roy (and Arnold on the list) are about how we did or did not use STC. That is a reasonable question for the TCG chair to ask and we have a well-reasoned answer for it. I would be willing to stand up and lead this discussion, but I really don't think we should present, just be ready to answer.

--
The question(s) from Francois are things we discusses at length within the TEG 
and we already agreed to comprimise on the functional representation (instead 
of operator syntax). I should know, I was one of the ones wanting operators 
and agreeing to the comprimise :-) I don't see the point of bringing this out 
again as it indirectly brings a bunch of other things onto the table. 
Specifically, if we go with functions, people can implement it inside the DB 
and we can (must) exclude anything that does not fit existing SQL syntax. If 
we go with operators, people must have an ADQL parser and once we 
admit/accept that position we are open to adding anything that does not fit 
standard SQL syntax.

--
On Francois' other issue with a spherical distance function taking simple 
numeric arguments, everyone will recall that in Cambridge we presented region 
without a point type and just numeric arguments. The one strong message that 
came out of that meeting was to have the point type (just as we now have it). 
If an existing DB has numeric columns, it means the query has some extra 
stuff to clarify, for example

DISTANCE ( POINT( a.coordsys, a.x, a.y), POINT( b.coordsys, b.ra, b.dec) )

and the service has to transform coordinate systems or throw an error.


** Side note: I think I said in the past that we could get rid of RECTANGLE 
completely without loss of functionality. We either need to do that or 
explicitly specify that it is a short-hand for a POLYGON (rather than a range 
of coord values). I would favour just removing it entirely and in the RFC 
response just say that we had discussed doing that but forgot to conclude :-)

sincerely,

-- 

Patrick Dowler
Tel/Tél: (250) 363-6914                  | fax/télécopieur: (250) 363-0045
Canadian Astronomy Data Centre   | Centre canadien de donnees astronomiques
National Research Council Canada | Conseil national de recherches Canada
Government of Canada                  | Gouvernement du Canada
5071 West Saanich Road               | 5071, chemin West Saanich
Victoria, BC                                  | Victoria (C.-B.)
Received on 2008-05-12Z19:53:19