update row if count(*) > n

Question

my DB has this structure:

ID | text | time | valid

This is my current code. I'm trying to find a way to do this as one query.

rows = select * from table where ID=x order by time desc;
n=0;
foreach rows{
    if(n > 3){
       update table set valid = -1 where rows[n]; 
    }
    n++
}

I'm checking how many rows exist for a given ID. Then I need to set valid=-1 for all rows where n >3;

Is there a way to do this with one query?

Answer 1

You can use a subquery in the WHERE clause, like this:

UPDATE table
   SET valid=-1
 WHERE (
         SELECT COUNT(*)
           FROM table tt
          WHERE tt.time > table.time
            AND   tt.ID = table.ID
) > 3

The subquery counts the rows with the same ID and a later time. This count will be three or less for the three latest rows; the remaining ones would have a greater count, so their valid field would be updated.

Answer 2

Assuming that (id,time) has a UNIQUE constraint, i.e. no two rows have the same id and same time:

UPDATE 
    tableX AS tu
  JOIN
    ( SELECT time
      FROM tableX
      WHERE id = @X                      -- the given ID
      ORDER BY time DESC
      LIMIT 1 OFFSET 2
    ) AS t3
    ON  tu.id = @X                       -- given ID again
    AND tu.time < t3.time 
SET
    tu.valid = -1 ;

Answer 3

update table 
   set valid = -1
 where id in (select id
                from table 
               where id = GIVEN_ID
            group by id
              having count(1) >3)

Update: I really like dasblinkenlight's solution because is very neat, but I wanted to try also to do it in my way, a quite verbose one:

  update Table1
     set valid = -1
   where (id, time) in (select id, 
                               time
                          from (select id,time
                                  from table1
                                 where id in (select id
                                                from table1
                                            group by id
                                              having count(1) >3)
                                -- and id = GIVEN_ID
                              order by time 
                                 limit 3, 10000000) 
                        t);

Also in SQLFiddle