I'm using ROW_NUMBER
and I'm trying to compare arr
in rn 1 to arr
in rn 2,3,4,etc
to see if they overlap. I can do this with a subquery / simple join. Is there a way that AVOIDS a join?
rn | id | job | arr |desired_result
---+----+-----+--------+---------
1 | 1 | 100 | {1,2} | {1,2}
2 | 1 | 101 | {2,3} | {1,2}
3 | 1 | 102 | {5,6,8}| {1,2}
4 | 1 | 103 | {2,7} | {1,2}
I made a dbfiddle
--USING JOIN
WITH a AS (
SELECT
ROW_NUMBER() OVER (PARTITION BY id ORDER by job) as rn
,*
FROM a_table
)
SELECT *
FROM (
SELECT id,arr
FROM a
WHERE rn = 1
) x
JOIN a
ON a.id=x.id
You can use first_value()
:
SELECT a.*, first_value(arr) over (partition by id order by job)
FROM a_table a;
row_number()
does not seem necessary.