[Shootout-list] OO (was Re: process creation & message passing)
Aaron Denney
wnoise@ofb.net
Tue, 19 Oct 2004 20:49:05 +0000 (UTC)
On 2004-10-19, Bengt Kleberg <bengt.kleberg@ericsson.com> wrote:
> >> otherwise we end up with test results like the current objinst. i
> >> think objinst is a failure.
> >
> > Why?
>
> because it claims to test ''the speed of object instantiation in OO
> languages'' and it has several entries of non-oo languages.
I would agree, but go further and say it's not useful because it has
several OO languages with different concepts of what OO is, and what
an object system provides. What an "object" consists of is too
vague. Now, you can wave your hands and say "well, the native object
for a given language is what people will use, so that's what we should
test."
That leaves the issue that several languages do not have a native
object system, but are perfectly capable of OO programing, either
with one of several packages or a small amount of dispatching glue
code. This naturally leads the submitter to include only what
they consider the bare essentials in their submission.
Depending on what features are considered key for OO, you'll get very
different object systems. Scheme has several, in fact too many to
easily list. Haskell[1] has one where you easily choose which features
(and what overhead) you want to include.
--
Aaron Denney
-><-
[1]
http://homepages.cwi.nl/~ralf/OOHaskell/
http://www.haskell.org//pipermail/haskell-cafe/2004-October/007138.html
http://www.haskell.org//pipermail/haskell-cafe/2004-October/007144.html