>>Bill,
>>
>>Exactly the justification for denormalizing. But denormalizing should not be taken lightly ro done before a fully normalized design is realized. You should use the calculations when printin reports so that regardless of anything that might cause a problem with synching the fields, the reports will alwys be consistent and driven from the details. This way accountants can easily make correcting entries.
>
>I don't understand your last sentence. What does correcting entries have to do with denormalizing data?
>
>I agree with you about only denormalizing AFTER creating a fully normalized design. But I do believe that being able to program calculations at the dbc level will cause the normalization rules to be rethought.

Bill,

Non-Normalized data creates the possibility for data inconsistencies. If a datum is stored in only one place it can never be inconsistent with itself. Once data appears in more than one place, inconsistency becomes a possibility. Inconsistency will prevent anyone from correcting inaccurate data because the client balance report says Joe owes me $1,200.00 and the invoice register says he owes me $1,500.00. So with that data, how do I make a correcting entry into the system. Which balance is the real balance? As database developers we aren't really concerned with the real balance, just providing reports that all show the same balance regardless of any tricks we may have used to get performance up or make coding easier.

I diasagree with the point of normalization rules being rethought. Regardless of any features of any specific data management product, normalizing data provides a number of advantages over not doing it. 1) new entities are discovered, 2) data redundancies are elimintated, 3) codependent fields are move to entities where they are not codependent, 4) parent attributes are moved from child tables into the parent tables, 5) data structures are made flexible and stable, 6) etc..

So, do I think that the new features in VFP allow it to be easier to handle non-normalized data, yes. Do I think that denormalizing data should ever be an easy decision, no!