Re: What databases have taught me

Neo wrote:
> > As some are pointing out now, OOP designs to not have to be hierarchical. However, outside of hierarchies, OO tends to lose its selling point. It is just a bunch of nodes (objects) with pointers to link them up, a big graph.
>
> I agree. In cases where data is highly structured, representing them
> with a RMDB provides many advantages. However in cases where data is
> highly unstructured, representing them with a RMDB can also become more
> difficult and starts to lose some of its advantages.

I am not sure I agree with that. I will agree that existing RDBMS brands make dynamism harder than it has to be, but that is only part of the issue.

Unstructured *anything* is going to be a bit difficult to use with any kind of automation. If "Name", "L_name", "Last_Name" are attributes all mixed together, for example, it will be tough to do any kind of coherant processing because a machine is not going to know that these are perhaps the same thing or related.

One can dump everything into an "attribute table" if there is no classification or "slot" for something, for example. Arbitrary graphs with arbitrary attributes can be created using a couple of many-to-many tables. If you really want a big blob of sloppy or inconsistently-labelled stuff, relational can model such. (Many on-server file systems I encounter are such messes, for example. Perhaps this is why the likes of Google exist.)

However, I am not saying that relational is best for every structuring need; just the majority of what I encounter in my domain.

-T- Received on Sat Jun 24 2006 - 01:04:21 CEST