On Mon, 2007-10-08 at 09:24 -0700, Joshua D. Drake wrote:
> > I haven't heard anyone say before that duplicate tuples were part of any
> > relational model.
> >
> > I'm not saying SQL is bad; it's certainly the best practical data
> > language we have.
> >
> > The problem I see is that it's the _only_ practical data language in
> > existence, and it is (in my opinion) imperfect.
>
> Jeff, that is fair but I must of course counter point. What language,
> regardless of implementation or purpose is perfect?
>
> And no Perl, is not perfect.
>
For something that's the only option available, and it is too far from
perfect in my opinion. There are many options for procedural
programming, functional programming, and object-oriented programming,
but only one practical option for a data language.
In SQL, "=" is both an assignment operator (e.g. UPDATE) and a
comparison operator (e.g. WHERE clause). There are nondeterministic
updates (if you have a FROM in your UPDATE) that produce different
results depending on the order rows are read from the heap (which might
depend on insertion order). There are all kinds of special words for
special cases, like "INTERVAL".
I think SQL could use some healthy competition. Perl may not be perfect,
but at least it's got plenty of viable alternatives.
Regards,
Jeff Davis