In a message dated 4/19/2004 1:34:50 AM Eastern Daylight Time, [EMAIL PROTECTED] writes:
> Now, would anyone care to post any experiences on performance of > implementing a star schema in a set of normalized U2 files and doing > drill down and roll up strictly with LIST, SORT, and its > options? Post > numbers, not guesses. Oooo I feel dirty just reading that. Why would you normalize U2 files if you're going to use LIST and SORT ? Isn't that the whole beauty of mv that we don't HAVE To normalize ? Yes I wrote a system where I could do what Pick historically called "cruising" and "double-clutching" ... that is link to another record, and "drill down" into details without much effort. Of course, looking back at my collection of 500 useless utilities, I was the only one who figured out how to use it well. I tried to teach my programmers, but, well .... So you pull up a list of say ORDERS, you select a link field say PART 110192-5 and it allows you to see the details of that PART record which then links to SALES.HIST or BILL.OF.MATERIAL or VENDOR or whatever. So you can go down, sideways, backup, go over there, jump here, and return to where you started, still in the original list you started with. The concepts are not horribly trickly, but the implementation is a "rhymes with switch". And you have to ensure that your linking fields are all built so the system can understand how to get the foreign record from the foreigh key i.e. your Tfile translates or TRANS functions have to be present. Will -- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users
