Re: SQL Implementation

joe.celko_at_northface.edu (--CELKO--) wrote in message news:<a264e7ea.0310031046.349c55c1_at_posting.google.com>...
> >> PL/I?: God, no! <<
>
> It compiled to over three times the size of a COBOL program to do the
> same job, and ALWAYS ran. Never mind that you wanted it to stop. The
> automatic type conversions could suddenly give you a payroll with
> complex numbers instead of a warning.
>

Didn't PL/I programmers get special keyboards with one extra key marked "Default" ? :)

> >> Algol-68? with pleasure <<
>
> No, no, no. Algol-60 was a pleasure. Algol-65 was a clean up.
> Algol-68 was so complex that there were only three compilers for it
> (one was the Royal Radar guys in teh UK and I don't remember the other
> two -- colleges, I think). I still have the specs for it; I still
> cannot understand the meta-meta-language they invented.
>

Algol-60 was almost a pleasure; but I can't say I can approve of any language where it's impossible to write a truly general swap procedure ! The Russians were pretty keen on Algol-68 - I'm sure somebody in St Petersburg did a Windows95 implementation, and I've even got a Linux interpreter for it - but I'm not sure whether anyone ever did a full-on implementation of truly general modes. (I lasted three pages of the official report - and I only kept plugging away to 3 so I could beat someone else's 2-and-a-half.)

> >> ADA? it's still alive and kicking <<
>
> Nope. The Ada mandate was killed on 1998 Oct 01.

Goodbye and good riddance to bad rubbish, says I.

> I was with AIRMICS
> when ADA was created and had to write code in it without a compiler.
> The thing was awful and the first compilers took a year longer than
> planned because of the complexity. As InfoTech put it, there was no
> way to build a kernel then add to it to get a full language compiler
> -- you had to create the entire language all at once.

And this was *by design* ! It was a declared intention that there should be no subsets of Ada - so why on earth did they throw every feature they could think of (and then some) into it ?

> A New York
> University built a compiler in SETL which had one error message and we
> played with that.
>

I know of a student Pascal compiler that had one error message - and one that doesn't bear repeating in polite company at that...

• Tony