Re: CREATE TABLE AS ... WITH DATA

[CRPence <CRPbottle@xxxxxxxxx>]

  But back to the original limitation... Would a cached plan be 
parameterized, if it could not be originally?  If not, then every 
separate invocation would have its own unique plan; cluttering the 
cache needlessly.  Using an INSERT INTO after the CREATE without 
data, at least allows both host variables and parameterized dynamic 
for which one copy is maintained.

Regards, Chuck

Vern Hamberg wrote:
> Yeah, the fully-dynamic approach does not keep an access plan 
> with the program. But in recent releases, there is a system-wide
> plan cache for statements run through the new engine, at least,
> and that is where these ad hoc plans are kept - but not over an
> IPL.
>
> Michael_Schutte@xxxxxxxxxxxx wrote:
> > Thanks, I figured that would be the case.
> >
> > Your "otherwise" suggestion... that's not good practice is it?
> > Because access paths are not saved with the program. It would
> > need to be determined with every execution.