Re: The stupidest design I ever saw

From: Neo <neo55592_at_hotmail.com>
Date: 6 Apr 2006 15:12:12 -0700
Message-ID: <1144361532.507695.27930_at_u72g2000cwu.googlegroups.com>


> > > table Nodes (id number, ... -- content)
> > > table Edges (tail number, head number, ...)
> >
> > How do I use the above schema to model the vehicle classes shown in the
> > RDF example given earlier? I still don't see it.
>
> No, you first please demonstrate that the above classification serves
> some useful purpose.

It may serve the purpose of allowing a large car dealership to narrow down the list of vehicles to offer based on a general category specified by a customer (ie mini-van, economy, economy/luxury, utility/hybrid, etc).

> Bruce Jacobs advocates that classification system for data management
> problems is a poor solution in general.

While I accept that implementation by a particular methodology may have limitations or problems (ie as in multiple inheritance in OOP), are you saying that classification, by itself, is a poor solution to data management? If so, why would human being rely on it so much? For example, if someone tells me thing X is a helicopter, it allows me to infer many characteristics. Next, if they tell me it is also a jet, it allow me to infer additional prominent characteristics and doubts some of the minor ones. For instance, I can now assume that it not only hovers but flies very fast, but less certain about it's cockpit/cabin environment. After learning more details, I might classify thing X predominantly in a new class (ie UFO?) but still keep it as a minor one in the old ones. I agree that various classes should not be arranged in a hierarchy (as in the earlier XML/RDF example) and that each thing simply belongs to 0 to many classes. In addition, human's data management system probably include weighting factors as to how strongly a thing belongs in a particular class and also for properties and methods associated with a class.

> Surprisingly, it's not easy to express this or parts explosion kind of
> query in XQuery (which is allegedly a superior tool for graph/tree
> related problems). You can compare XQuery solution ... with SQL ...

Thanks for presenting the two solutions. Received on Fri Apr 07 2006 - 00:12:12 CEST

Original text of this message