Exclusive OR condition needed for Check Constraint

I want to put a check constraint on a table to enforce the following
condition:
Of four columns A, B, C, D one must have a value and the other three must be
null.
A is a varchar column, the others are integer.
It seemed like I needed to create a bitwise Exclusive OR expression, with
some function to return true/false for each column having a value/Null.
The best I could do was this:
( isnumeric(len(A)) ^ isnumeric(B) ^ isnumeric(C) ^ isnumeric(D) ) = 1
This stops inserts when 0, 2 or 4 of the columns have values, and allows
inserts when 1 column has a value (which is correct). However it also allows
inserts when 3 of the columns have values. I have no idea why. Can anyone
fix the expression, or give me an alternative expression that fits the
requirements?
Thanks
Laurence Neville wrote:
> I want to put a check constraint on a table to enforce the following
> condition:
> Of four columns A, B, C, D one must have a value and the other three must be
> null.
> A is a varchar column, the others are integer.
> It seemed like I needed to create a bitwise Exclusive OR expression, with
> some function to return true/false for each column having a value/Null.
> The best I could do was this:
> ( isnumeric(len(A)) ^ isnumeric(B) ^ isnumeric(C) ^ isnumeric(D) ) = 1
> This stops inserts when 0, 2 or 4 of the columns have values, and allows
> inserts when 1 column has a value (which is correct). However it also allows
> inserts when 3 of the columns have values. I have no idea why. Can anyone
> fix the expression, or give me an alternative expression that fits the
> requirements?
isnumeric(len(A)) + isnumeric(B) + isnumeric(C) + isnumeric(D) = 1 ?
|||Try this:
... CHECK (
CASE WHEN A IS NOT NULL THEN 1 ELSE 0 END+
CASE WHEN B IS NOT NULL THEN 1 ELSE 0 END+
CASE WHEN C IS NOT NULL THEN 1 ELSE 0 END+
CASE WHEN D IS NOT NULL THEN 1 ELSE 0 END=1) ...
You could also consider changing your design. Perhaps you only need one
column.
David Portas
SQL Server MVP
|||Laurence,
Perhaps not the shortest nor efficient but this seems to work:
ALTER TABLE <TABLE> ADD CONSTRAINT <CONSTRAINTNAME> CHECK ((A IS NOT NULL
AND B IS NULL AND C IS NULL AND D IS NULL)OR (A IS NULL AND B IS NOT NULL
AND C IS NULL AND D IS NULL) OR (A IS NULL AND B IS NULL AND C IS NOT NULL
AND D IS NULL) OR (A IS NULL AND B IS NULL AND C IS NULL AND D IS NOT NULL))
HTH
Jerry
"Laurence Neville" <laurenceneville@.hotmail.com> wrote in message
news:%23%233KVUY2FHA.3592@.TK2MSFTNGP12.phx.gbl...
>I want to put a check constraint on a table to enforce the following
> condition:
> Of four columns A, B, C, D one must have a value and the other three must
> be
> null.
> A is a varchar column, the others are integer.
> It seemed like I needed to create a bitwise Exclusive OR expression, with
> some function to return true/false for each column having a value/Null.
> The best I could do was this:
> ( isnumeric(len(A)) ^ isnumeric(B) ^ isnumeric(C) ^ isnumeric(D) ) = 1
> This stops inserts when 0, 2 or 4 of the columns have values, and allows
> inserts when 1 column has a value (which is correct). However it also
> allows
> inserts when 3 of the columns have values. I have no idea why. Can anyone
> fix the expression, or give me an alternative expression that fits the
> requirements?
> Thanks
>
|||Yet another solution:
CHECK ( COALESCE(A, CAST( COALESCE(B,C,D) AS varchar(..)) ) = COALESCE(
CAST( COALESCE(D,C,B) AS varchar(..)), A) )
This solution approach would be cleaner if A had the same data type as
B, C and D. Then it would simply be:
CHECK ( COALESCE(A,B,C,D) = COALESCE(D,C,B,A) )
And I have to agree with David. If you need this check, then your data
model might not be properly normalized. In that case you might want to
review your design.
HTH,
Gert-Jan
Laurence Neville wrote:
> I want to put a check constraint on a table to enforce the following
> condition:
> Of four columns A, B, C, D one must have a value and the other three must be
> null.
> A is a varchar column, the others are integer.
> It seemed like I needed to create a bitwise Exclusive OR expression, with
> some function to return true/false for each column having a value/Null.
> The best I could do was this:
> ( isnumeric(len(A)) ^ isnumeric(B) ^ isnumeric(C) ^ isnumeric(D) ) = 1
> This stops inserts when 0, 2 or 4 of the columns have values, and allows
> inserts when 1 column has a value (which is correct). However it also allows
> inserts when 3 of the columns have values. I have no idea why. Can anyone
> fix the expression, or give me an alternative expression that fits the
> requirements?
> Thanks
|||Ah, forget this solution, it is no good :-(
Gert-Jan
Gert-Jan Strik wrote:[vbcol=seagreen]
> Yet another solution:
> CHECK ( COALESCE(A, CAST( COALESCE(B,C,D) AS varchar(..)) ) = COALESCE(
> CAST( COALESCE(D,C,B) AS varchar(..)), A) )
> This solution approach would be cleaner if A had the same data type as
> B, C and D. Then it would simply be:
> CHECK ( COALESCE(A,B,C,D) = COALESCE(D,C,B,A) )
> And I have to agree with David. If you need this check, then your data
> model might not be properly normalized. In that case you might want to
> review your design.
> HTH,
> Gert-Jan
> Laurence Neville wrote:
|||I went with David's solution because it is the easiest to interpret.
Mikito's solution also worked.
I know the table design is unusual and could be normalized. It is
deliberately this way to make certain queries perform faster (less joins to
make).
Thanks for so many quick replies!
"David Portas" <REMOVE_BEFORE_REPLYING_dportas@.acm.org> wrote in message
news:1130260200.705449.86860@.g49g2000cwa.googlegro ups.com...
> Try this:
> ... CHECK (
> CASE WHEN A IS NOT NULL THEN 1 ELSE 0 END+
> CASE WHEN B IS NOT NULL THEN 1 ELSE 0 END+
> CASE WHEN C IS NOT NULL THEN 1 ELSE 0 END+
> CASE WHEN D IS NOT NULL THEN 1 ELSE 0 END=1) ...
> You could also consider changing your design. Perhaps you only need one
> column.
> --
> David Portas
> SQL Server MVP
> --
>