Friends need sql help

Question

Dear Friends,
I have two selects running something like this:
select <column-names> from <table-names> where <conditions>
union
select <column-names> from <table-names> where <conditions>.
there are three outer-join conditions in the first select and
three outer joins in the second like:
c=l(+)[in the first select]
d=m(+)
e=n(+)
and
f=l(+) [in the second select]
g=m(+)
h=n(+)
The result here generates 52 rows which is correct.
Now the case is that I am removing the union clause from the above SQL
and converting it like this.
select <column-names[1]>+<column-names[2]> from <table-names[1]>+<table-names[2]> where <conditions[1]>+<conditions[2]>
but i get the error:
same table cannot be outerjoined to two tables in the same query because i now have this
in the new query :
c=l(+)[from the first select]
d=m(+)
e=n(+)
f=l(+) [from the second select]
g=m(+)
h=n(+)
if I remove the join like:
c=l[from the first select]
d=m
e=n
f=l(+) [from the second select]
g=m(+)
h=n(+)
the query executes but results into 160 rows too much of unneeded redundancy, which is wrong.
My question is that:"Is there any way to do this? I mean to remove the union and combine the two selects into one without affecting the result of the Query which should be 52 rows only (while not affecting the joins I think this should also be the case)".
Need your suggestions.
Thanks,
Vishal

Answer 1

Vishal,
What exactly are you trying to achieve? Removing/replacing the “union” is not really a goal.
You are comparing apples with oranges here.
select <column-names> from <table-names> where <conditions>
union
select <column-names> from <table-names> where <conditions>does one thing and …select <column-names[1]>+<column-names[2]> from <table-names[1]>+<table-names[2]> where <conditions[1]>+<conditions[2]>does another.
The results set cannot even be considered equivalent since one returns X columns and the other 2*X. Even if you ignore this aspect (how could you?) and compare the number of rows returned … first query returns N+M-D1 (where N is rows returned from sql1, M is rows returned from sql2 and D1 is some number determined by the implicit distinct done by the “union”) … the second query returns N*M-D2 (that is, the Cartesian product of sql1 and sql2 minus some D2 number if sql1 and sql2 have tables in common). N+M-D1 would be equal to N*M-D2 only by fluke.
Maybe, as Laurent put it, a concrete example with sample data and a proper stated goal are what you need to provide.