CODEWITHSUNDEEP
sqlsql-servert-sqlrandom
Lucas
Lucas

Reputation: 614

Using subquery to pull random value reveals same value every time

I want to add a random value to a column in a query using T-SQL for Microsoft SQL Server 2008 R2. To do this, I use a subquery and pull a random record. Every time I run the query, it pulls a new random record, but the value for every row is identical.

How can I get a new random value for each row?

Sample query:

SELECT column1,
    (SELECT TOP 1 column1 as c2 FROM Table2 ORDER BY NewID())
FROM Table1

Whereas table2 looks like

column1
value1
value2
value3

It always returns the same value for column2, instead of a new random record for each table.

First run:

column1, c2
1, value1
2, value1
3, value1

Second run:

column1, c2
1, value2
2, value2
3, value2

How can I get c2 (the subquery) to pull a new random record for each row?

Ideal scenario:

column1, c2
1, value2
2, value1
3, value2

Upvotes: 12

Views: 6558

Answers (2)

Martin Smith
Martin Smith

Reputation: 453426

You could try

SELECT T1.column1,
    (SELECT TOP 1 column1 as c2 FROM Table2 ORDER BY NewID(), LEN(T1.column1))
FROM Table1 T1

To force it to re-evaluate the selection each time. This will be very inefficient though. Also quite fragile (without the LEN it didn't work!) Would it meet your requirement to sort the T2 values randomly once (or possibly more if Table1 is larger than Table2) then join by row_number?

The main difference I can see with the methods is that your current method will allow the same random row to be selected multiple times my proposed method wouldn't.

Upvotes: 7

Thomas Li
Thomas Li

Reputation: 3338

Try:

SELECT column1,
    (SELECT TOP 1 column1 as c2 FROM Table2 WHERE Table1.column1 IS NOT NULL ORDER BY NewID())
FROM Table1

Upvotes: 19

Related Questions