Re: Creating a virtual OTLT (One True Lookup Table)

Mike Sherrill <MSherrillnonono_at_compuserve.com> wrote in message news:<cjbf60tvar425qhtncsdeg8ebevp1m17lq_at_4ax.com>...
> On Sat, 27 Mar 2004 16:54:34 -0500, "Laconic2" <laconic2_at_comcast.net>
> wrote:
>
> >But there are times when you need to lookup a code, but you have no idea of
> >what type of code it is, so you have no idea of which table to look in.
>
> When are those times?

remember, the OTLT problem is just a bill of material problem with lipstick. so, the proper sub-list of state codes might be determined by division/plant or region/product, depending on context. so, at design time, we know only that there are an unknown number of key columns (and the associated data types) into the state lookup table. we may know that today we must support four determinants, but unless we restrict the users to always use these four in the two combinations we define, we can't really design a useful lookup. if all we want is to *translate* code to description, then having multiple tables keyed only by the code value will work. but that's not a very interesting problem.

just as a bill of material, we must assemble an arbitrary part (the code and description list) from existing parts. viewed this way, parts is parts. Received on Thu Apr 01 2004 - 19:05:49 CEST