SQL：在列上返回具有最大值的多行

Question

You can find the problem from here .

When there are multiple people with the highest salary in each department, the following SQL command output the expected results - all employees with highest salary are output:

select d.Name as Department, e.Name as Employee, e.Salary as Salary
from 
(select Name, Salary, DepartmentId
from Employee
having (Salary, DepartmentId) in (
    select max(Salary), DepartmentId
    from Employee
    group by DepartmentId)
)e, Department d
where e.DepartmentId = d.Id

However, the following command only output one employee even if there are multiple employees with the same highest salary:

select d.Name as Department, e.Name as Employee, e.Salary as Salary
from 
(select Name, Salary, DepartmentId
from Employee
group by DepartmentId
having Salary = max(Salary)
)e, Department d
where e.DepartmentId = d.Id

Can anyone explain to me why the latter one does not work? Thanks!

Answer 1

The reason the second query doesn't work is because it's getting the max salary regardless of the department.

The way I would have done this, despite the valid first query (which is not using the standard JOIN syntax like it should), would be to first get the highest salary for each department like this:

SELECT departmentID, MAX(salary)
FROM employee
GROUP BY departmentID;

Once you have that, you can use a self join to only select the rows with that salary and department. This will include multiple people if they have the same salary:

SELECT e.*
FROM employee e
JOIN(
   SELECT departmentID, MAX(salary) AS maxSalary
   FROM employee
   GROUP BY departmentID) tmp ON tmp.departmentID = e.departmentID AND tmp.maxSalary = e.salary

---

For further explanation, consider just the inner subquery you have:

SELECT name, department, salary
FROM employee
GROUP BY department
HAVING salary = MAX(salary);

The first problem with this is that you are selecting rows that are not consistent within your group. For each department id, you have multiple name values and multiple salary values, which will be chosen arbitrarily when you do the grouping. In my example, and the first, you can see that the grouping explicitly pulls the (departmentID, salary) pair for the max salary in each department, which the second one does not.

Here is an SQL Fiddle example to visually see the differences between your second query and mine.