On 12/20/2013 11:11 AM, David Bicking wrote:
But isn't NULL and 0 a NULL?
I don't think it is. NULL really means unknown. The result of
"0 AND unknown" is not "unknown", it is 0. By contrast the
result of "0 OR unknown" really is unknown:
sqlite> SELECT (0 OR NULL);
null
"1 OR unknown" is not unknown, it is 1. And so on. To summarize:
sqlite> SELECT (0 AND NULL), (1 AND NULL), (0 OR NULL), (1 OR NULL);
0|null|null|1
Dan.
David
--------------------------------------------
On Thu, 12/19/13, Richard Hipp <d...@sqlite.org> wrote:
Subject: Re: [sqlite] Does not detect invalid column name when subexpression
optimized away
To: "R Smith" <rsm...@rsweb.co.za>, "General Discussion of SQLite Database"
<sqlite-users@sqlite.org>
Cc: amb...@math.bme.hu
Date: Thursday, December 19, 2013, 9:27 PM
On Thu, Dec 19, 2013 at
6:36 PM, RSmith <rsm...@rsweb.co.za>
wrote:
> my guess is
the optimiser pounces directly on the fact that (X and 0)
> will always be 0
>
Correct. The code is at
http://www.sqlite.org/src/artifact/962c2988?ln=556-559
sqlite3ExprAnd() gets called
to build the AND operator directly from the
parser, long before any name resolution has
occurred, and indeed before the
expression
has been fully parsed. As long as one side or the other of
the
AND operator does not have an egregious
syntax error, if the other side
evaluates to
FALSE then the whole expression is coded as a FALSE
literal.
That code was
added here:
http://www.sqlite.org/src/info/f9a7e179cbbeeab5
--
D. Richard Hipp
d...@sqlite.org
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
