Re: join query question

    I'm going to have to go look at my performance text now - you've got me wondering. What you say seems reasonable for tables of approximately equal size - but I have a ' avoid negation' clause in my head - and not just from blowing indexes. I'll have to dig.

You're right, maybe I am smelling 'vanilla' because somebody said it - not because it is in the air. I'll look .

jds

Brian Tkatch wrote:

> >The 'where not exists ' statement earlier can have detrimental connotations. Try to
> >avoid negation and base the seed on what is hopefully an indexed column ( empnum)
> >in both source tables.
>
> Why avoid negation? The original request was for everything in A after
> negating (from a) everything that is in B.
>
> Assuming empnum is indexed on an incrementing id, not exists should be
> able to traverse the index immediately to find where the id should be
> and know if it is or is not there. Am I missing some thing about hte
> was EXISTS works?
>
> Brian
Received on Fri Jun 22 2001 - 22:49:55 CDT