Re: Mixing OO and DB

From: S Perryman <q_at_q.com>
Date: Thu, 21 Feb 2008 15:58:02 +0000
Message-ID: <fpk72i$a2t$1_at_aioe.org>


Yagotta B. Kidding wrote:

> S Perryman <q_at_q.com> wrote in news:fpjf5b$rul$1_at_aioe.org:

>>What you are seeing is, after all this time, quite fundamental >>misunderstandings of what Liskov/Wings' work is actually about.

> The problem with the LSP so much beloved by the OOP crowd, apparently
> because the principle

There is no "principle" . You appear to have been sucked in like the rest. Re: http://groups.google.com/group/comp.object/msg/a2f843524b98522b

> gives an aura of theoretical respectability to the
> otherwise misguided and dull OOP enterprise

In your opinion.

> is that in general it is undecidable.

Lots of theoretical problem sets TP are in general undecidable.

But there are significantly subsets of any given TP that are in fact decidable, and not computationally expensive (NP-complete etc) . Which is useful. And also means ...

> In practical terms, it means that the type system is
> unable to check whether the LSP in a given piece of code is honored or
> violated rendering it pretty much a theoretical curiosity

... you are wrong (from the above) .

> In general, behavioural subtype checking is undecidable which means that
> those OOP program out there work only by accident !

Actually, it means *ALL* programs "work only by accident" . And if you claim this is not the case, you have contradicted your claim about behavioural subtype checking (irony eh) .

> For a more detailed treatment of the subtyping problem, you may want to
> visit Oleg Kiselyov's site :

> http://okmij.org/ftp/Computation/Subtyping/

Will have a look.

Regards,
Steven Perryman Received on Thu Feb 21 2008 - 16:58:02 CET

Original text of this message