Re: The IDS, the EDS and the DBMS

"mAsterdam" <mAsterdam_at_vrijdag.org> wrote in message news:413f41fd$0$37789$e4fe514c_at_news.xs4all.nl...
> > operations defined over them. Isn't this a form of integrity? Or am I
> > missing something?
>
> Not saying you are, checking the type of a value *is* somewhat
> special: You don't need any other time-variant data to do it.

Thanks for the clarification. Eventually, I'll get it. But by then it may be part of the "Stone age".

While we are on the subject....

How does Pascal do type checking on the value last assigned to a variable? It really doesn't. Pascal knows the type of the variable, and counts on the last assignment to have intended to store a value of the correct type. As long as Pascal performed the last assignment on behalf of the programmer, as is well.

But let's say I have an assembly language procedure that simply copies the first argument to the second, bit for bit, without regard to type? Call it, "cast(a; b)"

Now let's say I have

[code]
program pearls_before_swine (input; output); integer: a;
real: b, c;
begin
readln (a);
cast (a, b);
c := float (a);
writeln (a, b, c)
end.
[/code]

I know this is a perverse example. But I have a real interest. Consider a language of the Lisp family, that has variables ("atoms") that can be assigned values of any type. But you want the benfits of type checking. So you do type checking on values at execution time, instead of doing type checking on variables at compile time.

This is kind of far afield of DBMS systems. But if the DBMS, the PL, and the network are all integrated into one giant conceptual cosmos, it seems to me you have to deal with issues like this. Received on Wed Sep 08 2004 - 20:11:00 CEST