Comment on ADQL PR

[Yuji SHIRASAKI]

I have checked the compatibility between the ADQL and the 
standard SQL. Below is my comments I noticed.

1. ":"

   : is used to represent a host parameter in the standard SQL.

   The host parameter is used to exchange data between application
   program (C, C++, Java, ...) and SQL.

      select *
         from  :galaxy
         where :galaxy.ra between 20 and 30

   where ":galaxy" corresponds to the Java (or C, C++...) variable "galaxy".

   Question:

   Is the ":PhotoPrimary" in the following ADQL-s identified as a 
   host parameter ?

      select *
         from TWOMASS:PhotoPrimary
         where ...

2. "/"

   The expression of xpath should be delimited like "/Resource/Contact/Name"
   to distiguish the separator "/" in the xpath from the division operator 
   "/".

   If not, the following ADQL-s 

       Select /Resource/Contact/Name 
       From   Resource 
       where /Resource/Type like 庁ニcatalog庁ヌ

   will be parsed as

       Select /Resource / Contact / Name 
       From   Resource 
       where  /Resource / Type like 'catalog'

3. "[]"

   Accoring to the SQL92 and SQL99 specifiction, double quates should be 
   used to express the non-standard identifier as already discussed in 
   the recent voql list.

4. "INTO"

   The SQL standard uses SELECT INTO to represent selecting values into 
   scalar variables of a host program, rather than creating a new table.
   (http://www.postgresql.org/docs/8.0/interactive/sql-selectinto.html)

   So the following sql:

      Select g.* into VOS:/JHU/gal 
         from galaxy g
         where g.redshift > 3.5

   should be written as

      Create table VOS:/JHU/gal AS
         Select g.* 
            from galaxy g
            where g.redshift > 3.5

   And we need to standardize how to specify the table in the VO.
   Syntax of like "VOS:/JHU/gal" is not described in ADQL draft, is't it ?

5. "TOP"

   The order of the returned records should be specified when using the 
   "top" syntax, if not it is useless. The reason is that, if the order 
   is not specified, you will get an unpredictable subset of the query's 
   rows so a users cannot be sure from the return records that what they 
   want to get does exist in the database or not.

   This problem is also mentioned in the PostgreSQL user manual:

      LIMIT Clause section of:
      http://www.postgresql.org/docs/8.0/interactive/sql-select.html

6. "REGION"

   The syntax of region expression is not clearly defined.
   If we use STC syntax, STC document should be referenced.
   In what extent we incorporate the STC sould also be described.

7. "unit"

   The syntax of unit should be described or should refere to another
   document. It might be good to mandate the conversion among small 
   subset of units, like "deg", "arcmin", "arcsec"... and so on.
   Is it not neccessary to delimit the unit expression by "[]" or else ?

8. Explanation of select SQL syntax

   The syntax of select SQL that is actually used in ADQL should be 
   described in BNF and also human readable english sentense if possible.

   Section 8 "ADQL Grammer" describes the ANTLR grammar used to produce 
   the parser in C#, but I don't think that all the grammar is used in
   ADQL. Is it correct ?

---
Yuji SHIRASAKI, Research Fellow             +81-422-34-3579 (tel)
National Astronomical Observatory of Japan  +81-422-34-3840 (fax)
Osawa, Mitaka, Tokyo, Japan                 yuji.shirasaki at nao.ac.jp



From: Masatoshi OHISHI <masatoshi.ohishi at nao.ac.jp>
Subject: PRs (ADQL and SkyNodeInterface)
Date: Fri, 03 Jun 2005 17:53:17 +0900
Message-ID: <6.1.0.6.2.20050603174619.04b5feb0 at 133.40.4.4>

> Hi VOQL members,
> 
> Based on the agreements at the Kyoto InerOp meeting, I prepared two
> PR documents (ADQL-1.0 and SkyNodeInterface-1.0), and posted to
> the IVOA TWiKi page.
> 
> These documents are found at
>      http://www.ivoa.net/Documents/latest/ADQL.html
> and
>      http://www.ivoa.net/Documents/latest/SNI.html
> 
> I would like to ask you if these documents are OK to ask wider review.
> 
> I am going to decide around June 20th if it is OK to Request for Comments
> to interop at ivoa.net.
> 
> See below the formal procedure:
> 
>  From the IVOA documentation doc:
> : Next maturity level.  After a publication period of at least two weeks,
> : the chair of the Working Group that developed the Proposed
> : Recommendation may call for a formal Request for Comments (RFC).  The
> : RFC is sent to the widest possible IVOA distribution lists
> : (interop at ivoa.net) and announced on the website.  Distribution of the
> : RFC initiates a four-week public review period.  All comments submitted
> : during this review period must be posted publically (SIC) and responded to
> : publicly.  If the review identifies significant deficiencies in the
> : document, such that revisions must be undertaken beyond minor editorial
> : changes or where revisions require changes to software based on the
> : document, the document must be returned to the Working Draft status.
> 
> Cheers,
>     Masatoshi
> 
> 
>