Re: S.O.D.A. database Query API - call for comments

From: Tobias Brox <tobiasb_at_suptra.org>
Date: Sat, 21 Jul 2001 18:01:15 GMT
Message-ID: <9cs7on$3bp$1_at_tobiasb.invalid>


In comp.databases Per Schröder <per_at_nospam.mimer.se> wrote:
> If you consider Chris Date or Hugh Darwen to be powers within the
> relational world, you will find that they oppose the "class=relation" view.
> They propose "class=domain".

Reading the reffered interview, it seems that he has some strong arguments why. An object might be represented completely differently, but still have the same accessor methods. Also, any system should not differ between complex objects and simple integers.

Still, I find it a bit hard to accept that all data for a complex object should be stored within one column in one row. What if some weird part that's present in several airplanes of different kinds (classes?) cause a plane to fall down - then it's critical to find all airplanes having this particular part. You just can't do that if the airplane is one encapsulated object. Or have I misunderstood something?  

-- 
Tobias Brox - freelancer for hire!
Programming, system administration, etc
+47 98660706 / tobiasb_at_suptra.org
Received on Sat Jul 21 2001 - 20:01:15 CEST

Original text of this message