Re: Does Codd's view of a relational database differ from that ofDate&Darwin?[M.Gittens]

From: Jan Hidders <jan.hidders_at_REMOVETHIS.pandora.be>
Date: Tue, 21 Jun 2005 19:34:03 GMT
Message-ID: <LYZte.126579$Ww3.6933636_at_phobos.telenet-ops.be>


Alexandr Savinov wrote:
>
> I found that in many aspects FDM is very similar to COM, especially in
> query language. FDM is really very simple and rather natural. The main
> difference of COM from FDM is that in COM everything is about order. [...]
> As you say, it is possible to constrain it to a lattice. But
> the main problem is again in priorities:
> "I can constrain a model to a lattice (if I like it)"
> is completely different from
> "My model IS a lattice from the very beginning and nothing else".
> In particular, if I order relations in the RM as a lattice and then add
> a couple more constraints and features then I will get COM.

Agreed. But I think it is important that you should be honest about the trade-off here. You *do* introduce extra restrictions. Having said that, even though I don't know exactly what they are yet, I'm also quite convinced that there will be benefits because of this.

On another note, I still think it would be better for several reasons to piggy-back your ideas on existing data models. One reason is that you avoid alienating the referees by introducing yet another self-invented terminology for things we already have accepted terms for. Another reason is that such lattices are probably tied to different views that different groups of users may have upon the same data. Compare this to the way the UR approach is usually marketed. Rather than an alternative for the RM it is an interesting and useful way to define views that are easier to use, and there is no reason why you couldn't have several of such views upon the same underlying RM data. If you present your ideas in a similar way, they will be much more easier to accept.

  • Jan Hidders
Received on Tue Jun 21 2005 - 21:34:03 CEST

Original text of this message