Question

I have a table with three columns (apart from the id):

• topic
• user
• ip

Topic cannot be null, but the other two can. There are no unique restrictions.

So an example of data would be this:

What I want to get is the distinct combination of user and ip, and count the resulting rows grouped by topic:

EDIT:

Ok, the first popular answer is:

SELECT topic, COUNT(IP)
FROM MyTable
GROUP BY topic;

But I see you are ignoring the user column, why? Maybe my example was not good 🙂

Let’s add a new record:

Running the above query would give us the same result, but it is wrong:

SELECT DISTINCT topic, user, ip FROM MyTable;

It returns:

So in this case the totals would be:

Answer 1

SELECT Topic, COUNT(*) [Count]
FROM table
WHERE IP IS NOT NULL
GROUP BY Topic

Sample:

-- preparing the data
DECLARE @tbl TABLE (Topic int, [User] int, IP varchar(20))
INSERT INTO @tbl VALUES
(1,1,null),
(1,1,null),
(1,1,null),
(1,null,'127.0.0.1'),
(1,null,'127.0.0.1'),
(2,1,null),
(2,null,'127.0.0.1'),
(2,null,'127.0.0.1');

-- getting the resuls
SELECT Topic, COUNT(*) [Count]
FROM @tbl
WHERE IP IS NOT NULL
GROUP BY Topic

For the edited question:

;WITH TMP AS
(
    SELECT DISTINCT topic, [user], ip
    FROM @tbl
) 
SELECT Topic, COUNT(*)
FROM TMP
GROUP BY Topic