>>>
>>Tamar,
>>
>>Pretty much the same with me. As I mentioned to JimB, I've a bunch of intentionally de-normalized tables (because of the report requirements) which require multiple groupings (up to 20+ in one case). From strictly a maintenance standpoint, breaking the rules here is almost a necessity.
>>
>
>I know I joked with you about normalizing (in regards to this topic) a couple of weeks ago...

Who's to say what's "normal"?< vbg >

>Even if the data is normalized, you'll still end up de-normalizing to create cursors for the VFP report writer.

In one of my cases, there are three main files involved. Production, machines, and standards. Data from the latter two files must be pulled into the production file when the data is entered. Since machines and their associated attributes can change, and standards do change (usually quarterly) doing a three table join instead would return invalid results in some cases.

>So, yep, it would be a pain without SET ENGINEBEHAVIOR...

You got it.:-)