Re: Interesting article: In the Beginning: An RDBMS history

mAsterdam wrote:
> Marshall Spight wrote:
> >
> > The choice between a numerical vs. symbolic identification of
> > attributes is purely syntactic.
>
> I can not agree with that.
>
> First a frame of reference.
> Supporting levels, low to high:
> (fatic -) syntactic - semantic - pragmatic.

I don't understand this last paragraph at all. (What is "fatic?") But it seems to imply that syntactic is somehow "lesser" than semantic; I wouldn't agree with that. They are different things, but one is not "higher" than the other; both are necessary for language.

> Numerical vs. symbolic identification of attributes has
> consequences (i.e. differs at the pragmatic level.)

I'm not sure I understand this either. It seems to say that if something has pragmatic consequences, it can't be purely syntactic. If so, I strongly disagree.

Syntax is huge in what it can do. I have often underestimated its importance and its power, but I'm going to try not to do that anymore. I'm still not clear on what the boundaries of syntax are, but I'm working on it.

> Names rely on the existance of common semantics,
> numbers on form agreement (isomorphism)
> (Both only implied in practise most of the time,
> but hey! this is cd/t/ :-)
>
> This choice goes well beyond syntax.

Can you give a specific example?

Consider: we can write a relational language which contains join as an operator. We can write this language in such a way that the syntactic phase uses either numbers or names to identify relation attributes, and we can produce an abstract syntax tree that identifies attributes either by name or by position.

So how is the decision anything *other* than syntactic?

> > But wanting to have both at the same time, while it seems innocuous
> > enough, actually leads to the loss of important algebraic properties.
> > It seems like it can be reconciled, but I am convinced it can't, at
> > least not without some loss.
>
> Intuitively: Yep, I think it can't either. My guess: demarcating
> this loss will take some effort and time, though.

It certainly took me long enough!

> > I'm too tired to write up the substitutibility problem right now.
> > I'll just mention that I spent long time trying to devise a
> > syntax and semantics that would hold all the design value
> > of named attributes with all the notational convenience of
> > positional attributes, and I couldn't make it work. The
> > problems are subtle, but pernicious.
>
> I wish I had some nice words to encourage you to share
> those problems.

Um, well. Let's see.

An important property of a language is that you can bind a subexpression to a name and then rewrite the expression, substituting the name for the subexpression and get identical results.

Okay, I'm really going through a stupid phase. I'll have to defer again.

Marshall Received on Sun Apr 09 2006 - 11:07:07 CEST