Cart
[Home] [By Thread] [By Date] [Recent Entries]
[Bullard, Len] > Even if we can get the semantics sharable or just get them > all to agree that a boolean choice is not an int or is, we probably > built traps into our own relational table designs. > An issue is that many relational systems contain > a lot of system-specific metadata spread around throughout > the tables to data-drive the system itself. > Some bits are isolated into system tables, but > others are in the content tables. Think code systems for > picklists, user-extensible GUI features, etc. How about when the domain for a code table is built into the definition of a pick list that uses it (easy to do with a tool like Powerbuilder, for example)? That might be the only place where you could learn what the allowable strings and values really are. Ouch. Cheers, Tom P > Now the > question becomes, if you want to create a true content-centric > schema minus the GUI stuff, how well will that be > rehydratible even roundtripping in and out of the same system? > > Just [expletive deleted] the marrow out of the relational bones mechanically > isn't enough.
|
