Ken,

>OK - I understand your point now - sorry for the attitude, It's been a frustrating day and I unfairly unbloaded on you on that last post. I apologize.

Apology accepted...I think some of my statements made my point-of-view seem a bit wishy-washy, so I don't blame you for the teeny barb.

>I agree that there is a lot of disagreement on what lives where. I think the general philosophy in 3 tier - and the one I try to follow - is RI on the BE, BR in the middle. It becomes gray when trying to decide where to put what I call "conditional RI".

Nice summary. Yeah, conditional RI many times seems to be very hard-and-fast in the rules it uses, and so a database implementation seems like a good idea. And I would say that if you have a very good database design, it probably can all go there. One problem I had on a previous project was that we had a hard time accurately data modeling the A to Z of a printing job (I worked for R. R. Donnelley, world's largest printer at that time). Jobs had changeable bills of material, varied contractual details, and horrendously complex building and distribution instructions. Not to mention that we were trying to combine all types of print jobs (directory, tabs, catalogs, magazines, digests) into one model. Let's just say it was a mistake *grin*, and since we never got our data model right, I don't think we ever quite figured out how to get all the rules in right spots.

*smile*

Thanks for your insight into the SP side of things...

JoeK