More efficient double coalesce join alternative

Question

I have a procedure with a (slightly more complex) version of the below:

CREATE PROC sp_Find_ID (
    @Match1 varchar(10),
    @Match2 varchar(10)
) AS

DECLARE @ID int

SELECT @ID = ID
FROM Table1
WHERE Match1 = @Match1
    AND Coalesce(Match2,@Match2,'') = Coalesce(@Match2,Match2,'')

SELECT @ID ID

Essentially Match1 is a mandatory match, but Match2 is both optional on the input to the procedure, and on the table being searched. The 2nd match succeeds where the input and/or the table Match2 values are null, or where they're both the same (not null) value.

My question is: Is there a more efficient (or even more readable) way of doing this?

I've used this method a few times, and I feel slightly sullied each time (subjective dirtiness admittedly).

Answer 1

Seems simple to me? I must be missing something.. you dont need the Coalesce

SELECT @ID = ID
 FROM Table1
 WHERE Match1 = @Match1
    AND (
          (Match2 is null and  @Match2 is null)
           or
           @Match2=Match2
    )

SELECT @ID ID

Answer 2

I would have thought this should do it - providing that the @Match2 value will be NULL if it is optional.

CREATE PROC sp_Find_ID (
    @Match1 varchar(10),
    @Match2 varchar(10)
) AS

DECLARE @ID int

SELECT @ID = ID
FROM Table1
WHERE Match1 = @Match1
    AND Match2 = IsNull(@Match2, Match2)

SELECT @ID ID 

Answer 3

Don't know if this is any more preferable.

SELECT @ID = ID
FROM Table1
WHERE Match1 = @Match1
    AND ((Match2 = @Match2) OR Coalesce(Match2,@Match2) IS NULL)

Answer 4

Is there a more efficient (or even more readable) way of doing this?

The example you provided, using COALESCE/etc is non-sargable. You need to separate things so only what needs to be present in the query is run:

DECLARE @ID int

IF @Match2 IS NOT NULL
BEGIN

  SELECT @ID = t.id
    FROM TABLE1 t
   WHERE t.match1 = @Match1
     AND (t.match2 = @Match2 OR t.match2 IS NULL)

END
ELSE
BEGIN

  SELECT @ID = t.id
    FROM TABLE1 t
   WHERE t.match1 = @Match1

END

SELECT @ID ID

If you want this to occur in a single SQL statement, dynamic SQL is the only real alternative. I highly recommend reading The curse and blessing of dynamic SQL before reading further:

DECLARE @SQL NVARCHAR(MAX)
    SET @SQL = N' SELECT @ID = t.id
                    FROM TABLE1 t
                   WHERE t.match1 = @Match1 '

    SET @SQL = @SQL + CASE 
                        WHEN @Match2 IS NOT NULL THEN
                          ' AND (t.match2 = @Match2 OR t.match2 IS NULL) ' 
                        ELSE 
                          ' '
                      END

BEGIN

  EXEC sp_executesql @SQL,
                     N'@ID INT OUTPUT, @Match1 VARCHAR(10), @Match2 VARCHAR(10)',
                     @ID, @Match1, @Match2

END

Answer 5

Avoiding OR and ISNULL etc

• The EXCEPT bit returns no rows if either side IS NULL
• Match2 <> @Match2 means exclude non-NULL non-matching

Something like this

DROP TABLE dbo.Table1

CREATE TABLE dbo.Table1 (ID int NOT NULL, Match1 int NOT NULL, Match2 int NULL)
INSERT dbo.Table1 VALUES (1, 55, 99), (2, 55, NULL)

DECLARE @Match1 int = 55, @Match2 int

SELECT ID
FROM
    (
    SELECT ID FROM Table1 WHERE Match1 = @Match1
    EXCEPT -- @Match2 = NULL, match both rows (99, NULL)
    SELECT ID FROM Table1 WHERE Match2 <> @Match2
    ) foo

SET @Match2 = -1
SELECT ID
FROM
    (
    SELECT ID FROM Table1 WHERE Match1 = @Match1
    EXCEPT -- @Match2 = -1, match ID = 2 only where Match2 IS NULL
    SELECT ID FROM Table1 WHERE Match2 <> @Match2
    ) foo