comments on ADQL draft

[Roy Williams]

ADQL comments

(1) The standard as given is very simple in terms of coordinate frames 
etc -- assumes all handled elsewhere. What should a user do if they wish 
to query a table that is in galactic coordinates or B1950.0? What if the 
user wishes to write queries in other coordinate systems?

(2) The functions defined in the paper have poor extensibility because 
of these fixed fortran-like arguments. Why not plan for the future 
extension of the standard, and also make things more readable, with 
something like this:
POINT(frame="ICRS" equinox="1950" position2d="270.0, 22.0")

(3) The given document is difficult to read with its single-minded focus 
on bulding a BNF grammar. It needs examples to show the semantics and 
semiotics of the BNF statements. Without examples it is difficult for me 
to understand semantics: either how customers might respond, *or* how 
difficult it is to implement. How, for example is the LONGITUDE(POINT) 
used?
-- Can I say this:
    SELECT * from Table where LONGITUDE(POINT) between 270 and 285
If so, what kind of tables have POINT structures in them?
-- Can I say this to implement a SIAP?
    SELECT * from ImageTable where INTERSECT(POLYGON, SDSS_DR6_FOOTPRINT)
-- Can I say
    SELECT POLYGON from ImageTable
What kind of tables can be interpreted to have POLYGON attributes?

(4) What kind of table can I specify in the select clause? Can I use 
VOSpace, like this:
    SELECT * vos://nvo.caltech.edu/myTable.dat where INNER JOIN SDSS_DR6 
....
I ask this, because the DAL group is assuming that everyone will have a 
VOSpace that can provide temporary storage, and that crossmatches are 
done through ADQL queries. An alternative is for VOQL services to allow 
qualified users to upload tables, like CasJobs.

(5) There is already a specification of spatial regions (*) that is 
semantically identical to the STC standard, which is and IVOA 
Recommendation.
(*) http://www.ivoa.net/Documents/Notes/STC-S/STC-S-20071205.html, 
section 4.2
But the definitions in the ADQL document use different words and 
different meanings from this, (eg one says BOX and the other says 
RECTANGLE). Have the ADQL authors read the existing IVOA standards? If 
so, what has been inherited from the STC standard into the ADQL?


-- 

California Institute of Technology
626 395 3670