ADQL/XML non compliance with SQL?

[Tony Linde]

Nearest BNF I could find included:

<search condition> ::=
      <boolean term>
    | <search condition> OR <boolean term>

<boolean term> ::=
      <boolean factor>
    | <boolean term> AND <boolean factor>

<boolean factor> ::=
    [ NOT ] <boolean test>

<boolean test> ::=
    <boolean primary> [ IS [ NOT ]
          <truth value> ]

<boolean primary> ::=
      <predicate>
    | <left paren> <search condition> <right paren>

Which would indicate that we ought to have <LeftParen> and <RightParen>
elements in the language for explicit setting of precedence.

Cheers,
Tony. 

> -----Original Message-----
> From: owner-voql at eso.org [mailto:owner-voql at eso.org] On 
> Behalf Of Ray Plante
> Sent: 09 February 2004 18:13
> To: voql at ivoa.net
> Subject: Re: ADQL/XML non compliance with SQL?
> 
> 
> On Mon, 9 Feb 2004, Martin Hill wrote:
> 
> > Hi Ray,
> > 
> > As you put it yourself, should we be imposing these 
> precedence rules?
> 
> We should try to follow the SQL standard, whatever that is.
> 
> > My
> > understanding of SQL in practice is that statements with 
> many sequential ANDs 
> > (or sequential ORs) are resolved at the database in the 
> most efficent manner, 
> > rather than blindly left to right.  Clive Page probably 
> knows more about this.
> 
> I believe this is true.  As I understand it, the DB may not 
> strictly evaluate left to right for efficiency purposes; 
> however, the final result must match according to the 
> precedence rules.  
> 
> > If this is so, I guess this means the discussion has changed to 
> > deciding
> > between 
> > whether we 'stick rigorously to the SQL syntax' or 
> 'interpret for practical 
> > cases'.
> 
> If we need predictable behavior across all DBs and the SQL 
> standard does 
> not provide that, then we may need to apply the latter.
> 
> cheers,
> Ray
>