"David Allardyce" <[EMAIL PROTECTED]> writes:
> ... If they had defined uniqueness as "all rows
> must be distinct" then two all null rows would violate the uniqueness
> constraint. Not the behavior I want at all.
Er, why not? You're essentially arguing that the UNIQUE constraint
should treat n
I don't mean to re-hash an argument that has been debated-to-death before,
but I can't help myself...
> > However, shouldn't any values that are not NULL violate the constraint
if
> > the same values exist already?
>
> No. Postgres is conforming to the SQL standard in this. SQL92 saith
> in sec
"David Allardyce" <[EMAIL PROTECTED]> writes:
> However, shouldn't any values that are not NULL violate the constraint if
> the same values exist already?
No. Postgres is conforming to the SQL standard in this. SQL92 saith
in section 4.10:
A unique constraint is satisfied if and only
It appears that Postgres will allow any INSERT, despite a multiple-column
constraint, if any of the values INSERTed are NULL. If I read the included
excerpt correctly (there are like three negatives in the second sentence,
sheesh :) ), multiple NULL values for a column are acceptable or, in other
