Using ROWID returns less rows than if use distinct in query

Question

Hi,
I am writting  the following query to get distinct purchase order number from a specific received date. this returns me total 620 rows.:
select  distinct pha.segment1 from apps.po_headers_all pha, apps.RCV_TRANSACTIONS rcv
where 1=1
and trunc(rcv.transaction_date) between to_date(:P_FROM_DATE,'YYYY/MM/DD HH24:MI:SS') and to_date(:P_TO_DATE,'YYYY/MM/DD HH24:MI:SS')
and pha.po_header_id=rcv.po_header_id
and pha.type_lookup_code='STANDARD'
and pha.authorization_status ='APPROVED'
and pha.cancel_flag='N'
if i write the query using ROWID to get the distinct Purchase order number its giving me 580 rows:
select  pha.segment1 from apps.po_headers_all pha, apps.RCV_TRANSACTIONS rcv
where 1=1
and trunc(rcv.transaction_date) between to_date(:P_FROM_DATE,'YYYY/MM/DD HH24:MI:SS') and to_date(:P_TO_DATE,'YYYY/MM/DD HH24:MI:SS')
and pha.po_header_id=rcv.po_header_id
and rcv.rowid=(select max(rowid) from apps.RCV_TRANSACTIONS rcv where rcv.po_header_id=pha.po_header_id)
and pha.type_lookup_code='STANDARD'
and pha.authorization_status ='APPROVED'
and pha.cancel_flag='N';
Can anyone tell me why? I want to use this ROWID query in a valueset in oracle apps.

Answer 1

That's probably because some purchase orders have rcv transactions out of the range (p_from_date, p_to_date). You should include the condition on the dates in the (select max(rowid)...) subquery as well. And more generally, the first query seems like a better/simpler choice than the second one.
Hope it helps.