Re: Complex (for me) SQL question

Hi Dan,

LATERAL joins make that simple:
--------------------------------------------------------------------
with CTE_DUPLICATES as (
select PRPRDN
from HPRDFA
group by PRPRDN
having count(*) > 1
)

select DETAIL.*
from CTE_DUPLICATES D

cross join lateral (
select H.*
from HPRDFA H
where H.PRPRDN = D.PRPRDN
order by H.PRDATER
fetch first 2 rows only
) as DETAIL

order by DETAIL.PRPRDN, DETAIL.PRDATER
--------------------------------------------------------------------

LATERAL joins => learn 'em, love 'em, use 'em...:)

The ROW_NUMBER() function can also handle that task nicely.

Mike

date: Mon, 18 Jan 2016 15:33:51 -0500
from: Dan <dan27649@xxxxxxxxx>
subject: Complex (for me) SQL question

I have a table with duplicate (by part number) rows. The following query
lists only those rows for which duplicates exist:

with sum1 as (
select PRPRDN, count(*) as count1 from HPRDFA
group by PRPRDN
order by PRPRDN)
select * from HPRDFA d
where exists (select * from sum1 s where s.PRPRDN = d.PRPRDN and
s.count1 > 1)
order by PRDATER

I need to have this query list only the first two records for each part
number (field PRPRDN), ordered by PRDATER (process date). I'm thinking
somewhere along the lines of "fetch first 2 rows only", but this will
return only two records for the entire query, not two records for each part
number.

- Dan