Re: An object-oriented network DBMS from relational DBMS point of view
From: Dmitry A. Kazakov <mailbox_at_dmitry-kazakov.de>
Date: Tue, 13 Mar 2007 21:29:18 +0100
Message-ID: <8561pqgk3rlj.14qe02xknxg7w.dlg_at_40tude.net>
>
> I don't warry about correctness. no one Turing machine can prove that
> another machine will stops.
>
> why? implementation of vtbl can be done extreamelly fast.
Date: Tue, 13 Mar 2007 21:29:18 +0100
Message-ID: <8561pqgk3rlj.14qe02xknxg7w.dlg_at_40tude.net>
On 13 Mar 2007 11:04:47 -0700, Dmitry Shuklin wrote:
> On 12 мар, 20:55, "Dmitry A. Kazakov" <mail..._at_dmitry-kazakov.de>
> wrote:
>> Self-recursive types is IMO a bad idea, because the recursion can turn >> infinite, so that the correctness would become undecidable.
>
> I don't warry about correctness. no one Turing machine can prove that
> another machine will stops.
This is exactly why correctness is an issue. If you can't verify it, better ensure it per design.
>> Of course, that could be circumvented with dynamic typing and a >> lot of dynamic casts, but this is not the way to go, IMO.
>
> why? implementation of vtbl can be done extreamelly fast.
It is not about performance. Then existence of dispatching tables is not an equivalent of dynamic typing. The difference between static and dynamic typing with respect to dispatching is essentially about whether the latter may fail at run-time. In a feasible statically typed system it never fails, per types construction.
-- Regards, Dmitry A. Kazakov http://www.dmitry-kazakov.deReceived on Tue Mar 13 2007 - 21:29:18 CET