联结表中的额外user_id为简单起见

Question

I have 3 tables:

foo(id, ..., user_id)
bar(id, ...)
foo_bar(foo_id, bar_id)

I need to select a row from 'foo_bar' and check if the record belongs to a specific user. Is it good practice to add a 'user_id' column to 'foo_bar' table so that I could run a query:

// 1 query only 
$foobar = SELECT * FROM foo_bar WHERE foo_id = ? AND bar_id = ? AND user_id = ?

instead of first querying 'foo' table to check if 'foo' record belongs to user and then running a second query to select the 'foo_bar' record? Something like this:

// 2 queries
$userId = SELECT user_id FROM foo WHERE foo_id = ?;
if ($userId === $userIdToCompareTo) {
    $foobar = SELECT * FROM foo_bar WHERE foo_id = ? AND bar_id = ?
}

Does this make sense or is this usually handled another way with a more complex query?

Answer 1

No, you should not. You should look up the value using a join :

select . . . 
from foo_bar fb join
     foo f
     on fb.foo_id = f.id
where f.user_id = ?;

A key principle of data modeling is that you should store data in only one place. That is, if you store the user_id in foo_bar , then what happens if the user_id changes? Or if the foo record is deleted?

There are some situations where you would want to replicate the user id. In particular, if the user id does change, you might want to capture the user_id on foo_bar at the moment when the record is created. I only bring this up because it is important to understand principles and to understand when they don't necessarily apply.