Gustav Brock wrote: > So this sounds as poor design. I disagree. > The ID is and should be regarded as a "meaningless" value > with one purpose only: to identify the object (record). That is irrelevant.
> This is not specific to Cach� but to any neatly designed > database scheme. You missed her point. > If you need another ID to match something in the outside world, > do follow the suggestion of Kevin and create a separate ID for > that purpose - you will never regret it. That won't solve the problem either. The "problem" is the late lookup. Suppose I give you data about a Person. You think it is a new Person. But when it comes time to save, the program discovers that the gender, birth date, first name, and home telephone number match another Person. The program asks you whether this existing Person is your man. Sure enough, it is, just a minor spelling error in the last name kept you from finding the person before. Now what do you do? Do you want to abandon everything and make the user re-enter?
