sql select only first 3 rows from second table - sql

Question

I only need 3 records from child table.
CREATE TABLE project
(prj_code int , prj_name varchar(255) );
CREATE TABLE task
(id int , prj_code int, task_name varchar(255) );
INSERT INTO project VALUES (1,'myproj1');
INSERT INTO project VALUES (2,'myproj2');
INSERT INTO task VALUES (1,1,'testing');
INSERT INTO task VALUES (2,1,'prod');
INSERT INTO task VALUES (3,1,'prod2');
INSERT INTO task VALUES (4,1,'testing2');
INSERT INTO task VALUES (5,2,'mytask1');
INSERT INTO task VALUES (6,2,'mytask2');
Please see the fiddle
EDIT:
Want to pivot table to retrieve the 3 smallests tasks Ids in columns for each project

Answer 1

This code does the trick. You need to use RANK to select the 3 smallests IDs and then pivot table:
WITH tasks AS (
select
A.prj_code,
prj_name,
RANK() OVER (PARTITION BY A.PRJ_CODE ORDER BY ID ) RANK,
ID,
TASK_NAME
from task A
inner join project B
on A.PRJ_CODE = B.PRJ_CODE
order by A.prj_code)
select PRJ_CODE, prj_name, sum(T1) T1, sum(T2) T2, sum(T3) T3
from tasks A
PIVOT (max(ID) for RANK IN ('1' as T1 ,'2' as T2 ,'3' as T3 )) B
GROUP BY PRJ_CODE, prj_name
OR
use the code bellow to have the tasks names
select PRJ_CODE, prj_name, max(T1) T1, max(T2) T2, max(T3) T3
from tasks A
PIVOT (max(TASK_NAME) for RANK IN ('1' as T1 ,'2' as T2 ,'3' as T3 )) B
GROUP BY PRJ_CODE, prj_name
SQLFiddle

Answer 2

This is a hideous solution, but I think it illustrates a way to accomplish this. I don't assume any pivoting capabilies, not knowing what version of Oracle you're running.
with tasks as (
select
a.prj_code, a.prj_name,
b.id, b.task_name,
row_number() over (partition by a.prj_code order by b.id) as rn
from
project a
join task b on
a.prj_code = b.prj_code
), pivotdata as (
select
prj_code, prj_name,
case when rn = 1 then task_name end as task_1,
case when rn = 2 then task_name end as task_2,
case when rn = 3 then task_name end as task_3
from tasks
)
select
prj_code, prj_name,
max (task_1) as task_1, max (task_2) as task_2,
max (task_3) as task_3
from pivotdata
group by
prj_code, prj_name
Output:
Prj Code Prj Name Task 1 Task 2 Task 3
1 myproj1 testing prod prod2
2 myproj2 mytask1 mytask2