16

I know that you can't compare a value to NULL and expect a result without adding something like in the following code...

SELECT
    *
FROM 
    A INNER JOIN 
    B ON A.ID = B.ID
WHERE
    A.STRING <> B.STRING OR (A.STRING IS NULL AND B.STRING IS NOT NULL) OR (A.STRING IS NOT NULL AND B.STRING IS NULL) OR 
    A.DT <> B.DT OR (A.DT IS NULL AND B.DT IS NOT NULL) OR (A.DT IS NOT NULL AND B.DT IS NULL) OR 
    A.B <> B.B OR (A.B IS NULL AND B.B IS NOT NULL) OR (A.B IS NOT NULL AND B.B IS NULL) OR 
    A.NUM <> B.NUM OR (A.NUM IS NULL AND B.NUM IS NOT NULL) OR (A.NUM IS NOT NULL AND B.NUM IS NULL) 

My question is:

Is there a more elegant way to test for changed values across the two tables where either one could be null?

Solution needs to work uniformly across the data types.

Here is the code to set up the test tables...

CREATE TABLE A
(
    ID INT IDENTITY(1,1) NOT NULL,
    STRING VARCHAR(20) NULL,
    DT DATETIME NULL,
    B BIT NULL,
    NUM INT NULL
)

CREATE TABLE B
(
    ID INT IDENTITY(1,1) NOT NULL,
    STRING VARCHAR(20) NULL,
    DT DATETIME NULL,
    B BIT NULL,
    NUM INT NULL
)


INSERT INTO A (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO A (STRING, DT, B, NUM) VALUES (NULL, '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO A (STRING, DT, B, NUM) VALUES ('TEST', NULL, 0, 23)
INSERT INTO A (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', NULL, 23)
INSERT INTO A (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, NULL)
INSERT INTO A (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO A (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO A (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO A (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO A (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)


INSERT INTO B (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO B (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO B (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO B (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO B (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO B (STRING, DT, B, NUM) VALUES ('STAGE', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO B (STRING, DT, B, NUM) VALUES ('TEST', '2555-11-11 00:00:00.000', 0, 23)
INSERT INTO B (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO B (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 23)
INSERT INTO B (STRING, DT, B, NUM) VALUES ('TEST', '2012-03-16 16:39:04.893', 0, 999)
24

You can use this approach from Paul White's article Undocumented Query Plans: Equality Comparisons

SELECT * 
FROM   A 
       INNER JOIN B 
         ON A.ID = B.ID 
            AND EXISTS(SELECT A.* 
                       EXCEPT 
                       SELECT B.*) 
  • This is great, I tested it out and it seems to behave exactly as I need it to. Thanks very much. – GWR Mar 16 '12 at 20:28
  • 1
    Drat it you beat me to it! Though I have to confess you taught me this method. It's so beautiful. :) – ErikE Mar 16 '12 at 20:47
  • Martin, are you using this approach to compare the tables that are too large for Data Compare? – A-K Mar 17 '12 at 23:09
0

This is a global approach so i didn't rewrite a corrected query from question. I just want to share a useful point maybe help, So below script will check if the Column1 and Column2 is equal, Additionally used Convertion to VARBINARY to compare in case sensitive and you can remove it if not necessary.

--c1 = Length of Column1
--c2 = Length of Column2

ISNULL(NULLIF(CONVERT(VARBINARY(cl), LTRIM(RTRIM(Column1))),
              CONVERT(VARBINARY(c2), LTRIM(RTRIM(Column2)))),
       NULLIF(CONVERT(VARBINARY(c2), LTRIM(RTRIM(Column2))),
              CONVERT(VARBINARY(c1), LTRIM(RTRIM(Column1))))) IS NULL

You can change the end of expression to IS NOT NULL for checking unequal condition.

Hope this help.

3

Standard SQL, supported in SQL Server 2005 and better:

WITH A_MINUS_B 
     AS
     (
      SELECT * 
        FROM A
      EXCEPT 
      SELECT *
        FROM B
     )
SELECT * 
  FROM A_MINUS_B AS T 
       JOIN B ON T.ID = B.ID;

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.