On 1-10-2018 18:18, Kovalenko Dmitry wrote:
Hello,
Firebird 3.0.4
Could anybody (Adriano?) confirm, that next result is corrected:
select
null or true as c4__null_or_true, /* TRUE */
null or false as c5__null_or_false, /* NULL */
null and true as c6__null_and_true, /* NULL */
null and false as c7__null_and_false /* FALSE */
from rdb$database
Just – yes or no :)
Yes, and verified against SQL:2016.
You can reason about if you substitute TRUE and FALSE for NULL (or
UNKNOWN as for booleans that is interchangeable) and see how that
influences the result.
For (UNKNOWN or TRUE), you can reason that whether UNKNOWN is replaced
with TRUE or FALSE, the result is TRUE, so UNKNOWN is irrelevant for the
result.
But for (UNKNOWN or FALSE), you can't because if replaced with TRUE, the
result is TRUE, but with FALSE the result is FALSE. That means that
UNKNOWN decides the result, so the result is UNKNOWN.
Same for AND:
(UNKNOWN and FALSE), whether you replace with TRUE or FALSE, the result
is FALSE.
And for (UNKNOWN and TRUE), UNKNOWN decides the result: if you replace
with TRUE, the result is TRUE, with FALSE, the result is FALSE.
Mark
--
Mark Rotteveel
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
