Re: Relational URI (was: Re: Relational Calculus URIs?)
Date: Wed, 21 Apr 2004 12:05:28 -0500
"Laconic2" <laconic2_at_comcast.net> wrote in message
> So, if there were such a thing as URI, what would it be?
> How about "Uniform Resource Identifier"?
> The trouble with URL is that it's an address. Basically, a pointer. And
> when a resource moves from one "location" to another, all the pointers to
> the old location get left behind, and invalidated.
> Worse, when a resource gets replaced by a different resource, with a
> different identity, but the same location, the pointer acts like it's
> valid, although it actually should not be.
golly this sounds familiar
You are familiar with URI's, right?
Thanks for recognizing my existence on this list (it wasn't easy to have a voice here as some might recall) and, I agree -- I suspect that URL thing will never catch on. If it does, it simply won't work over time. No users will put up with missing links. It is bound to tank, uh, soon. Having a single "key" to a document is just a bad idea, especially if there is any intelligence in that key (such as domain name). It turns the model into one of Functions rather than the most more friendly extension to functions -- relations.
Let's watch and see whether the WWW "functions" and di-graph model or the
RDBMS model fades sooner.
P.S. I just filled in for two courses for two weeks for a Calculus professor on a paternity leave, not having taught calculus for 20 years. I saw how much of the introductory courses are devoted to functions rather than the superclass of relations (an exception being conic sections). If we want simplicity in data modeling, using functions seems like a good plan, so I think I'll promote it ;-) Received on Wed Apr 21 2004 - 19:05:28 CEST