Richard Huxton wrote:
PFC wrote:
NULL usually means "unknown" or "not applicable"
Andrew Sullivan wrote:
Aaaargh! No, it doesn't. It means NULL. Nothing else.
If it meant unknown or not applicable or anything else, then
SELECT * FROM nulltbl a, othernulltbl b
WHERE a.nullcol = b.nullcol
would return rows where a.nullcol contained NULL and b.nullcol
contained NULL. But it doesn't, because !(NULL = NULL).
(a == b) <=> ( (a -> b) AND (b -> a))
| a | b | a->b | b->a | a==b |
|----|-----|------|------|------|
| F | F | T | T | T |
| F | T | T | F | F |
| F | U | U | U | U |
| T | F | F | T | F |
| T | T | T | T | T |
| T | U | U | T | U |
| U | F | U | U | U |
| U | T | T | U | U |
| U | U | U | U | U |
Ergo, (UNKNOWN = UNKNOWN) is UNKNOWN. Similarly for (UNKNOWN != UNKNOWN).
Where NULL differs is that (NULL = NULL) is FALSE, and (NULL != NULL) is FALSE.
The similarity is that with NULL, SQL is not exactly saying (NULL = NULL) is
FALSE so much as that it's not TRUE.
NULL follows Zen-valued logic, not 3-valued, and that seems somehow
appropriate to me.
--
Lew
