Re: Objects and Relations

On Jan 31, 6:13 am, "JOG" <j..._at_cs.nott.ac.uk> wrote:
> On Jan 30, 7:12 pm, "Marshall" <marshall.spi..._at_gmail.com> wrote:
>
> > On Jan 30, 10:18 am, "Neo" <neo55..._at_hotmail.com> wrote:
>
> > > > I wish I had a decent general purpose relational language I could prototype this stuff in. Starting in March I hope to be able to put some decent amount of effort into that. SQL is so clunky!
>
> > > So what is a proper relational expression for the string bob?
>
> > #:int, c:char
> > { (0, 'b'), (1, 'o'), (2, 'b') }
>
> > But I would expect a syntactic shortcut so you could just
> > write:
>
> > "bob"
>
> > as is customary in a variety of languages.
>
> Indeed, and hopefully the OP will recognize how similar this is to the
> actual contents of the string object itself.

Sure.

> I'd also note that while
> RDBMS are obviously not traditionally setup to handle this sort of
> thing (in terms of physical implementation and optimization, as
> opposed to the underlying theory), there is no reason a programming
> language could not be constructed that represented a string datatype
> as such.

When you say "represented" I presume you mean logically not physically.

I have no problem separating current RDBMS from RM. I understand that RM is orthogonal to concerns of persistence, concurrency control, atomicity, durability etc. Lightweight support for RM would probably be very useful.

Nevertheless IMO RM is a poor choice for modelling strings and expressing string processing algorithms. I agree that this is unsubstantiated conjecture. In keeping with this theory newsgroup it is made without regard for what tools currently exist (or don't exist). Received on Wed Jan 31 2007 - 04:26:30 CET