We really should not encourage the piddling focus of things like these
benchmarks. It's a bit like comparing restaurants by how fast you can
finish a dinner at each one: it's really missing the point.
On 12/5/07, Raul Miller <[EMAIL PROTECTED]> wrote:
>
> On Dec 5, 2007 5:46 PM, Randy MacDonald <[EMAIL PROTECTED]> wrote:
> > You can get the result using the same method outlined in the benchmarks:
>
> In my opinion, benchmarks which micromanage implementation, rather
> than specify outcomes, can only be meaningful when considering different
> implementations of the same underlying language. And, even there, their
> only value is in determining how well some aspect of that language fits
> some resource constraint.
>
> Extending this kind of benchmark across languages seems to be somewhat
> like trying enforce the use of latin verb conjugation rules in korean.
>
> Put differently, in contexts where these benchmarks ("implementation is
> enforced") are meaningful across languages, we are probably really talking
> about things like dialects -- different implementation of more or less
> exactly
> the same underlying symbolism -- rather than significanlty different
> languages.
>
> Put differently, if the underlying language took a "one at a time
> specification"
> and turned that into something like J's L2,|.L3, it would do well at this
> kind
> of benchmark. So what do we gain from forbidding L2,|.L3?
>
> --
> Raul
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
--
Devon McCormick, CFA
^me^ at acm.
org is my
preferred e-mail
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
