问题
I was practicing self join and here's a thing I do not understand in writing query.
I have a table 'employee'
The employee table contains three records.
+-----+---------------+------------+
| id | employee | manager_id |
+-----+---------------+------------+
| 1 | Ola | NULL |
| 2 | Ahmed | 1 |
| 3 | Tove | 1 |
+----------+----------+------------+
Last column manager_id refers to the first column id making Ola manager of Ahmed and Tove.
If I write the query like
SELECT emp.employee as NAME, manager.employee as MANAGER
FROM employee as emp, employee as manager
WHERE emp.id = manager.manager_id
Result makes Ahmed and Tove Manager. Whereas
SELECT emp.employee as NAME, manager.employee as MANAGER
FROM employee as emp, employee as manager
WHERE manager.id = emp.manager_id
Makes it correct, could anyone please explain?
回答1:
Self join is like a inner join where two or more instances of same table are joined together through a common data type column/field. Such join(inner join) gives the common rows as result, based on the joining condition.
The employee table contains three records. In this case,
employee as emp:
+-----+---------------+------------+
| id | employee | manager_id |
+-----+---------------+------------+
| 1 | Ola | NULL |
| 2 | Ahmed | 1 |
| 3 | Tove | 1 |
+----------+----------+------------+
employee as manager:
+-----+---------------+------------+
| id | employee | manager_id |
+-----+---------------+------------+
| 1 | Ola | NULL |
| 2 | Ahmed | 1 |
| 3 | Tove | 1 |
+----------+----------+------------+
Now First case: Lets try this to understand the difference:
SELECT emp.*, manager.*
FROM employee as emp, employee as manager
WHERE emp.id = manager.manager_id
+-----+---------------+------------+-----+---------------+------------+
| id | employee | manager_id | id | employee | manager_id |
+-----+---------------+------------+-----+---------------+------------+
| 1 | Ola | NULL | 2 | Ahmed | 1 |
| 1 | Ola | NULL | 3 | Tove | 1 |
+----------+----------+------------+----------+----------+------------+
See, emp.id = manager.manager_id . Thus, emp.employee as NAME is giving rows of Ola from first table & manager.employee as MANAGER is giving rows of Ahmed & Tove from the second table.
Now Second case: Lets try this to understand the difference:
SELECT emp.*, manager.*
FROM employee as emp, employee as manager
WHERE manager.id = emp.manager_id
+-----+---------------+------------+-----+---------------+------------+
| id | employee | manager_id | id | employee | manager_id |
+-----+---------------+------------+-----+---------------+------------+
| 2 | Ahmed | 1 | 1 | Ola | NULL |
| 3 | Tove | 1 | 1 | Ola | NULL |
+----------+----------+------------+----------+----------+------------+
See, manager.id = emp.manager_id . Thus, emp.employee as NAME is giving rows of Ahmed & Tove from first table & manager.employee as MANAGER is giving rows of Ola from the second table.
回答2:
Writing WHERE emp.id = manager.manager_id does not make much sense, because manager (or the row you want to display as a manager) does NOT have a manager_id. I.e you must start with emp.manager_id because you want to list employees and to each manager_id of that employee you want to list the corresponding manager.
回答3:
manager.employee will give you the correct name only if you JOIN on manager.id
which only happens in the second query.
The alias of the table which uses id in the ON condition MUST be the same
which uses employee column,so manager.id->manager.employee.
回答4:
For each row, manager_id refers to the employee's manager. This is described by your second query. Here you match emp.manager_id to the employee in the joined table. The asserts that the relationship exists because of the id column -- in other words, Ola's manager is anyone with a manager_id of 1. In this case, both Ahmed and Tove have a manager_id of 1, so they both match. HTH.
来源：https://stackoverflow.com/questions/22274129/understanding-self-join