"Jonathan Lewis" <jonathanjlcomp.demon.co.uk> wrote in message news:<atpg31$ckd$1$8300dec7news.demon.co.uk>...you're right, of course. i guess i have come to expect magic from> I like the 'nothing unusual there' line.
> Oracle has unnested a subquery, and
> recognised a partition view - that's a
> pretty good feat of optimisation.
apparently, it does. mine is set to false.> Do you have parameter
> partition_views_enabled = true ?
> Very recent versions of Oracle recognise
> partition views even when this parameter
> is set to false - I didn't think the behaviour
> went back to 22.214.171.124.
it's not so much that i want partition elimination, just a predicate> Oracle will automatically try to push joins
> into UNION ALL views and do partition
> elimination provided it recognises the UNION ALL
> view to be a partition view - which means that
> all the tables in the UNION ALL have exactly the
> same structural definitions. This means that
> all the tables must have the same columns,
> and types, in the same order, and all indexes
> must match across all tables - and the view must
> be equivalent to:
> select * from table1
> union all
> select * from table2
push that will allow use of the indexes on the underlying tables. a
range scan (+ table access by rowid) on both indexes would be fine. i
was under the impression that this was achievable, even with a view of
union (ALL or otherwise, but without identical tables). is there no
way to get this to happen?
thanks for your answer.
> Jonathan Lewis
> Coming soon a new one-day tutorial:
> Cost Based Optimisation
> (see [url]http://www.jlcomp.demon.co.uk/tutorial.html[/url] )
> Next Seminar dates:
> (see [url]http://www.jlcomp.demon.co.uk/seminar.html[/url] )
> ____England______January 21/23
> The Co-operative Oracle Users' FAQ