Very Computer

I decided to fix a design flaw in a system I'm working on to split a
database table into 2. I created a view which joined them back together in a
way that was indistinguishable from the original.

Most of the system worked unchanged, however a few of the most horrendously
complex SQLs with some old fashioned syntax

(LEFT INNER JOIN (SYMBOLS INNER JOIN

etc... involving 9 or more tables produced bad results when the table was
replaced with the view.

Is there any "known problem" I should be aware of concerning this? Has
anyone experienced this & found an easy fix?

I really don't want to attempt to rewrite those SQLs because of the testing
load that would add to the project. So the alternative is to go back to the
old inefficient table structure. I would really appreciate any suggestions
which would allow me to avoid doing that.

Thanks.

NEIL

Hi,

Can you post the old table structure, the new one and one of the queries
that misbehaves (with enough DDL to get it parsed)?
The Version of DB2 is helpfull, too :-)

Would be interesting to see where it goes of...
One reason that comes to mind is that DB2 could now be sharing thw view
rather than redoing the join everytime. This could lead to system temp
tables and failure to push down some predicates.

Cheers
Serge

--
Visit the DB2 UDB and DB2 Connect Online Support site at:
http://www.ibm.com/software/data/db2/udb/winos2unix/support