Question

I have this query :

SELECT
    p.name,
    COUNT(DISTINCT t.keyTask) AS totalTasksCount,
    SUM(DISTINCT CASE WHEN t.keyPriority = 24 AND (t.keyState = 16 OR t.keyState = 17) THEN 1 ELSE 0 END) AS highPriorityTasksCount,
    SUM(DISTINCT CASE WHEN t.keyState = 16 OR t.keyState = 17 THEN 1 ELSE 0 END) AS activesTasksCount,
    SUM(DISTINCT t.estimatedDuration) AS estimatedDuration,
    SUM(TIMESTAMPDIFF(SECOND,wp.start,wp.end)) * 1000 AS workedDuration
FROM projects_projects p
LEFT JOIN projects_tasks t
    ON p.keyProject = t.keyProject
LEFT JOIN projects_workPeriods wp
    ON wp.keyTask = t.keyTask
LEFT JOIN common_organizations o
    ON o.keyOrganization = p.keyOrganization
LEFT JOIN common_users uc
    ON uc.keyUser = p.keyUserCreator
LEFT JOIN common_users uu
    ON uu.keyUser = p.keyUserUpdater
GROUP BY
    p.keyProject
ORDER BY
    highPriorityTasksCount DESC,
    activesTasksCount DESC,
    p.updated DESC,
    p.name;

But the result fields highPriorityTasksCount and activesTasksCount returns 0 or 1 which is normal with this query. I was wondering if there is any way to make DISTINCTROW work on row and not on case result value for these fields without subquery ?

Current result :

p.name,
totalTasksCount,
highPriorityTasksCount,
activesTasksCount,
estimatedDuration,
workedDuration

'Project 1',  '4', '1', '1',  '14400000',  '15300000'
'Project 2', '48', '1', '1',  '84600000', '503100000'
'Project 3',  '6', '1', '1', '108000000',        NULL
'Project 4',  '4', '1', '1',  '25200000',  '30600000'
'Project 5',  '5', '1', '1', '226800000',  '39600000'
'Project 6',  '2', '0', '1',        NULL,  '10800000'
'Project 7',  '9', '0', '1',        NULL,  '36900000'

Expected result :

'Project 1',  '4', '1', '1',  '14400000',  '15300000'
'Project 2', '48','20', '2',  '84600000', '503100000'
'Project 3',  '6', '1', '1', '108000000',        NULL
'Project 4',  '4', '4', '2',  '25200000',  '30600000'
'Project 5',  '5', '5', '1', '226800000',  '39600000'
'Project 6',  '2', '0', '1',        NULL,  '10800000'
'Project 7',  '9', '0', '1',        NULL,  '36900000'

EDIT :

Modified query with subqueries since there is no way, help about optimization would be appreciated, this one is working :

SELECT
    p.name,
    COUNT(DISTINCT t.keyTask) AS totalTasksCount,
    /* OLD SUM(DISTINCT CASE WHEN t.keyPriority = 24 AND (t.keyState = 16 OR t.keyState = 17) THEN 1 ELSE 0 END) AS highPriorityTasksCount, */
    (
        SELECT
    SUM(CASE WHEN st.keyPriority = 24 AND (st.keyState = 16 OR st.keyState = 17) THEN 1 ELSE 0 END)
    FROM projects_tasks st
    WHERE
    st.keyProject = p.keyProject
) AS highPriorityTasksCount,
    /* OLD SUM(DISTINCT CASE WHEN t.keyState = 16 OR t.keyState = 17 THEN 1 ELSE 0 END) AS activesTasksCount, */
    (
        SELECT
    SUM(CASE WHEN st.keyState = 16 OR st.keyState = 17 THEN 1 ELSE 0 END)
    FROM projects_tasks st
    WHERE
    st.keyProject = p.keyProject
) AS activesTasksCount,
    SUM(t.estimatedDuration) AS estimatedDuration,
    SUM(TIMESTAMPDIFF(SECOND,wp.start,wp.end)) * 1000 AS workedDuration
FROM projects_projects p
LEFT JOIN projects_tasks t
    ON p.keyProject = t.keyProject
LEFT JOIN projects_workPeriods wp
    ON wp.keyTask = t.keyTask
LEFT JOIN common_organizations o
    ON o.keyOrganization = p.keyOrganization
LEFT JOIN common_users uc
    ON uc.keyUser = p.keyUserCreator
LEFT JOIN common_users uu
    ON uu.keyUser = p.keyUserUpdater
GROUP BY
    p.keyProject
ORDER BY
    highPriorityTasksCount DESC,
    activesTasksCount DESC,
    p.updated DESC,
    p.name;

Answer 1

without subquery.

No, you will have to use a subquery, or there’s something I don’t know about MySQL.