Performance - direct joins vs joining datasets

Question

Which is better way to build query and which would give better performance
select * from A, B
where A.id=B.id
or select * from
(select * from A), (select * from B)
join on A.id =B.id
Essentially, join directly on tables or get results and join the datasets. This might too simplistic a example above

Answer 1

Hi,
927740 wrote:
Which is better way to build query and which would give better performance
select * from A, B
where A.id=B.idThat's the best way to do it.
or select * from
(select * from A), (select * from B)
join on A.id =B.idThis won't work. Table A is only in scope in the first (unnamed) in-line view, and B is only in scope in the second (unnamed) in-line view, so neither can be refererenced in the main query. You would need to give table aliases to the in-line views if you need to reference a column name (like id) that exists in both.
Essentially, join directly on tables or get results and join the datasets. This might too simplistic a example aboveJoin the tables directly when you have a choice.
There's no point in using
FROM    (SELECT * FROM a)rather than just
FROM    aIf there is a difference in performance, the simpler way will probably be faster, but the optimizer will probably use the same plan either way. The more complicate way will just confuse whoever has to debug and maintain the code.