Question

Could anyone explain to me the difference between the SQL statements below? I can see there is a difference but I can’t nail down the exact conditions which can make them produce different results. By the way, I’m thinking the distinct clause doesn’t make a difference on the user.id field since all the ids are already unique. The purpose of the query is to count the number of unique (non-empty) last names. If the last name is empty, then count as unique.

I suppose the general case for this problem would be the use of an aggregate function within a case-when statement.

Count within Case-When:

SELECT 
    (case when (substr(u.name,40,40) <> '                                        ')
        then count(distinct(substr(u.name,40,40)))
        else count(u.id) 
    end) as "LAST_NAME",
FROM 
    users u
GROUP BY
    substr(u.name,40,40)

Case-When within Count:

SELECT 
    count (distinct case when (substr(u.name,40,40) <> '                                        ')
        then substr(u.name,40,40)
        else to_char(u.id)
    end) as "LAST_NAME",
FROM 
    users u
GROUP BY
    substr(u.name,40,40)

Answer 1

If user.id is a PRIMARY KEY, these queries are identical semantically, though they are likely to produce different execution plans.

They will return 1 for all non-empty last names, since you are counting distinct values of the group-by expession within its group, which, by definition, will be exactly one.

For empty last names, the first query will essentialy return COUNT(u.id) and the second one will return COUNT(DISTINCT TO_CHAR(u.id)), which, given that u.id is unique, is the same.

I believe you need to remove GROUP BY from the second query:

SELECT  count (distinct case when (substr(u.name,40,40) <> '                                        ')
        then substr(u.name,40,40)
        else to_char(u.id)
    end) as "LAST_NAME",
FROM 
    users u