Join tables without losing the question

The idea

A join relates rows from two sources using a stated condition. An inner join retains matching pairs; a left join also retains each unmatched left row with NULL values on the right. Before writing the query, describe the expected relationship: one-to-one, one-to-many, or many-to-many. A parent with three matching children produces three joined rows, which matters when counting or summing.

Filter placement can change the meaning of an outer join. A right-side condition in ON restricts eligible matches while preserving unmatched left rows. Moving that condition into WHERE may remove those rows because their right-side values are NULL. Select explicit columns and use aliases to clarify ownership. Test a parent with several matches, one with only excluded matches, and one with none.

WITH teams(id,name) AS (VALUES(1,'A'),(2,'B')),
 tasks(id,team_id,done) AS (VALUES(10,1,0),(11,1,1))
SELECT teams.name, tasks.id
FROM teams LEFT JOIN tasks
 ON tasks.team_id=teams.id AND tasks.done=0
ORDER BY teams.id;

name | id
A    | 10
B    | NULL

The ON condition admits only unfinished tasks while the left join still preserves team B, which has no matching task.