Strange situation with joins!!!

Question

Hi,
I am having a weird situation with joins. I have two tables "EMPLOYER" and "ADDR". Table "ADDR" has columns "ADDRESS_ID" and "ADDRESS1".
Table "EMPLOYER" has two columns, "BUS_ADDRESS_ID" and "MAIL_ADDRESS_ID" which are left outer joined to the "ADDR" table on "ADDRESS_ID". I have also created an alias for "ADDR" to get both addresses.
The logical table Employer has two LTS, one with "ADDR" and one with "ADDR1". (I do not want to create one LTS with both the tables as left outer joins, that will create other problems)
When I query EmployerID,EmployerName,ADDR.ADDRESS1,ADDR1.ADDRESS1, I get the correct results, but the query is very slow.
I looked at the log and there is no second left outer join* at all, but the results are correct!!!. I have attached the log sequence.
-------------------- Logical Request (before navigation):
RqList distinct
EMPLOYER.EMPLOYER ID as c1 GB,
EMPLOYER.EMPLOYER NAME as c2 GB,
EMPLOYER.PRIMARY ADDRESS 1 as c3 GB,
EMPLOYER.MAILING ADDRESS 1 as c4 GB
DetailFilter: APPLIED EMPLOYER.EMPLOYER ID = 7501
OrderBy: c1 asc, c2 asc, c3 asc, c4 asc
+++USER1:1750000:1750004:----2009/03/04 15:22:22
-------------------- Sending query to database named DEV (id: <<2378603>>):
select T30717.EMPLOYER_ID as c1,
T30717.EMPLOYER_NAME as c2,
T44205.ADDRESS_1 as c3,
T30717.MAIL_ADDRESS_ID as c4
from EAPP_EMPLOYER T30717, ADDR T44205
where ( T30717.BUS_ADDRESS_ID = T44205.ADDRESS_ID (+) ) and ( T30717.EMPLOYER_ID = 7501 )
order by c4
+++USER1:1750000:1750004:----2009/03/04 15:22:22
-------------------- Sending query to database named DEV (id: <<2378624>>):
select T43657.ADDRESS_1 as c1,
T43657.ADDRESS_ID as c2
from
ADDR T43657
order by c2
+++USER1:1750000:1750004:----2009/03/04 15:22:30
-------------------- Query Status: Successful Completion
+++USER1:1750000:1750004:----2009/03/04 15:22:30
-------------------- Rows 1, bytes 160 retrieved from database query id: <<2378603>>
+++USER1:1750000:1750004:----2009/03/04 15:22:30
-------------------- Physical query response time 0 (seconds), id <<2378603>>
+++USER1:1750000:1750004:----2009/03/04 15:22:30
-------------------- Rows 0, bytes 0 retrieved from database query id: <<2378624>>
+++USER1:1750000:1750004:----2009/03/04 15:22:30
-------------------- Physical query response time 0 (seconds), id <<2378624>>
+++USER1:1750000:1750004:----2009/03/04 15:22:30
-------------------- Physical Query Summary Stats: Number of physical queries 2, Cumulative time 0, DB-connect time 0 (seconds)
+++USER1:1750000:1750004:----2009/03/04 15:22:30
-------------------- Rows returned to Client 1
+++USER1:1750000:1750004:----2009/03/04 15:22:30
-------------------- Logical Query Summary Stats: Elapsed time 8, Response time 7, Compilation time 0 (seconds)
Has anyone come across this situation? I have used the same concept in the same RPD for another model, but that used Oracle 9i as the datasource. This one uses Oracle 8i.
Thanks
rkingmdu

Answer 1

1) Have you used dfferent joins for the 2 versions of the addr table (origianal and alias).
i.e. one joins on bus_addr_id to employer and
the other joins to mail_addr_id to employer - *+{color:#ff0000}YES{color}+*
2) do both have left outer join specified? - *+{color:#ff0000}YES{color}+*
3) can you run employee, business_addrss query once
then run employee, mail_addrss query once separately, to see if the individual queries give correct results? - *+{color:#ff0000}YES, I get the correct results{color}+*
4) also check if the business_addrss and mail addrss in the logical employer table are mapped to different LTSs - *+{color:#ff0000}YES, they are{color}+*