MERGE INTO and CROSS JOIN

Question

Hello Guys,
I have a bit of trouble understanding the MERGE I have written here.
All the code executes without errors. As far as I see,
the below MERGE SELECT is a Cartesian product/cross join and I remember that is not
good. The ON (join condition) seems to be only for the merge into table (T_CLAIMS_BB) and
the table that comes from the select (d).
merge into T_CLAIMS_BB t
using
SELECT sch_vt.SCH#,vert.CNO,vers.SubCNo,regv.ReturnNo,regv.SndReturnNo,sch_vt.MainProd,sch_vt.Prod
FROM mainschema.TCONTRACT vert, mainschema.TVERSCONTRACT vers, commonschema.VC_REGVEREINB regv, myschema.T_CLAIMS_VT sch_vt
) d
ON (
t.sch# = d.sch# AND
SUBSTR(UPPER(t.BOOKINGTEXT),1,2) = d.SndReturnNo AND d.SubCNo IS NULL
) -- Join
when matched then
update
set t.CNO = 'a',
t.ReturnNo = 'a',
t.MainProd = 'b',
t.Prod = 'c';
I wonder now, what is the advantage of using MERGE in my case - if there is one.
Finally I want to demonstrate how the whole thing would look without a MERGE:
SELECT vers.* FROM TVERSCONTRACT vers
LEFT OUTER JOIN myschema.T_CLAIMS_BB sch ON sch.SubCNo = vers.SubCNo
LEFT OUTER JOIN commonschema.VC_REGVEREINB regv ON sch.SCH# = regv.SCH# AND SUBSTR(UPPER(sch.BUCHTEXT),1,2) = regv.SndReturnNo AND sch.SubCNo IS NULL
LEFT OUTER JOIN myschema.T_CLAIMS_VT sch_vt ON sch.SCH# = sch_vt.SCH#

Answer 1

Hi,
metalray wrote:
Hello Guys,
I have a bit of trouble understanding the MERGE I have written here.
All the code executes without errors. As far as I see,
the below MERGE SELECT is a Cartesian product/cross join and I remember that is not
good.Cross joins are good for some purposes. In practice, they're useful in only 1 or 2 queries out of every 1000.
Using the old join notation, it was very easy to get a cross join by mistake. Given how rarely cross joins are needed, the vast majority of cross joins (using old notation) were not good, but that was the fault of the programmers and the notation, not an indication that cross join itself was bad.
The ON (join condition) seems to be only for the merge into table (T_CLAIMS_BB) and
the table that comes from the select (d).Exactly!
You can have separate ON conditions inside the USING sub-query, if you need them.
>
merge into T_CLAIMS_BB t
using
SELECT sch_vt.SCH#,vert.CNO,vers.SubCNo,regv.ReturnNo,regv.SndReturnNo,sch_vt.MainProd,sch_vt.Prod
FROM mainschema.TCONTRACT vert, mainschema.TVERSCONTRACT vers, commonschema.VC_REGVEREINB regv, myschema.T_CLAIMS_VT sch_vt
) d
ON (
t.sch# = d.sch# AND
SUBSTR(UPPER(t.BOOKINGTEXT),1,2) = d.SndReturnNo AND d.SubCNo IS NULL
) -- Join
when matched then
update
set t.CNO = 'a',
t.ReturnNo = 'a',
t.MainProd = 'b',
t.Prod = 'c';
I wonder now, what is the advantage of using MERGE in my case - if there is one.I don't see any advantage, but I suspect the statement, as written, is not doing what it's supposed to. If someone made the mistake of using MERGE when UPDATE would be better, that might be the least of the mistakes in this statement.
Finally I want to demonstrate how the whole thing would look without a MERGE:It looks like the MERGE statement above is equivalent to:
UPDATE     t_claims_bb
SET     cno          = 'a'
,     returnno     = 'a'
,     mainprod     = 'b'
,     prod          = 'c'
WHERE     UPPER (SUBSTR (bookingtext, 1, 2)) IN
          SELECT  returnno
          FROM     commonschema.vc_regvereinb
AND     EXISTS
          SELECT  1
          FROM     mainschema.tverscontract
          WHERE     subcno     IS NULL
AND     EXISTS
          SELECT     1
          FROM     mainschema.tcontract
AND     EXISTS
          SELECT     1
          FROM     myschema.t_claims_vt
;Again, I suspect that the MERGE state,ment above is not doing what it was meant to do, or may coinncidentally be getting the right results on a small sample set. Many of the tables named in the USING clause of the MERGE statement do'nt seem to have much of a role in this problem.
SELECT vers.* FROM TVERSCONTRACT vers
LEFT OUTER JOIN myschema.T_CLAIMS_BB sch ON sch.SubCNo = vers.SubCNo
LEFT OUTER JOIN commonschema.VC_REGVEREINB regv ON sch.SCH# = regv.SCH# AND SUBSTR(UPPER(sch.BUCHTEXT),1,2) = regv.SndReturnNo AND sch.SubCNo IS NULL
LEFT OUTER JOIN myschema.T_CLAIMS_VT sch_vt ON sch.SCH# = sch_vt.SCH#What is this query? Perhaps you meant to have something like this in the USING clause of the MERGE.
I hope this answers your question.
If not, post a little sample data (CREATE TABLE and INSERT statements, relevant columns only) for all the tables involved, and the results you want from that data.
The sample data should show what the tables are like before the MERGE (or UPDATE), and the results will be the contents of the changed table after the MERGE.
Explain, using specific examples, how you get those results from that data.
Always say what version of Oracle you're using.