| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
Home -> Community -> Usenet -> comp.databases.theory -> Re: OO and relation "impedance mismatch"
Gene Wirchenko wrote:
> Kenneth Downs <firstinit.lastname_at_lastnameplusfam.net> wrote:
>
> [snip]
>
>>The current arbitrary feature for this case, which did not seem worth >>mentioning in that post, is that you can specify a suffix to affix to the >>columns in the child table. So for two foreign keys to the same table one >>is suffixed "_parent" and one suffixed "_child". This makes me really >>uneasy though, I keep thinking "salesperson1", "salesperson2", etc...> bad design, but that is not necessarily the case.
>
> Perfectly legitimate. That is not a 1NF violation. It may be
Agreed on all examples. But the mechanism allows 2NF violations as well, with no way to distinguish between them. I have been asking myself if there is some way to allow the legitimate cases while preventing 2NF violations. The salesperson example I keep giving as a problem follows this pattern:
The reason they did the 2NF violation at point 2 was simple, it was easier.
The only idea I've explored in thought experiments to any degree is the idea that if it were easier to maintain 2NF, more people would do it (or at least more people who work for me). Most frameworks are OK on creating a new child table and UI code for maintaining it, the trick seems to be getting that 1 column over to the child table during upgrade, and then making it look like nothing happened for people who don't need this mod. As you can imagine, I have not pursued that too much just yet, it is a lower order of problem.
-- Kenneth Downs Use first initial plus last name at last name plus literal "fam.net" to email meReceived on Thu Oct 07 2004 - 07:34:50 CDT
![]() |
![]() |