I think this, but untested:
"Field_1" AS ID,
"Field_2" AS ID_REF,
"Field_1"&"Field_2" as Unique_ID
SQL SELECT *
Or just join the tables together in SQL, again untested and almost certainly with syntax errors, and doing an inner join which may not be what you wanted:
,date(date#(Task_Date,'DD-MMM-YYYY')) as Task_Date
SELECT A.Field_1 as ID
, A.Field_2 as ID_REF
, A.Field_3 as Task_Date
, B.blah_blah_blah as blah_blah_blah
FROM ODBC.TABLE_NAME_A A
, ODBC.TABLE_NAME_B B
WHERE A.Field_3 >= '01-DEC-2008'
AND B.Field_1 = A.Field_1
AND B.Field_2 = A.Field_2
the first sugestion not at all, the second doesn't seem to work either.
The second load sequence doesn't reference the first set of data that has been collected. When I do a limited load of 10 lines, the second script gets a different 10 lines/ID's that probably do exist in the original tables, but that isn't what I want.
Any other ideas please?
I have attached the file so you can see in detail what I am trying to do.
ROBS_TEST.qvw 188.5 K
I'm not sure how much more I can help. Your two choices, as I see them, are some sort of where exists(...) or doing the desired join in SQL. If the syntax I gave you is incorrect (likely), or is being translated to the real example incorrectly, or if a limited load causes its own problems, I can't test it, so can't really fix it. I could put together an example with inline loads, but the syntax on those differs some from the syntax on SQL, and I can't mimic the second method at all, so it probably wouldn't be particularly helpful.