[Shootout-list] Directions of various benchmarks

[Bengt Kleberg]

skaller wrote:
...deleted

> So I accept completely your argument, but not as an argument
> in favour of 'same way' requirements, but as an argument
> in favour of throwing out most tests than lead to similar results
> for all languages, given 'same thing' arguments (that is,
> tests that simply require a correct result no matter
> how it is obtained).

1 i agree that same thing is better than same way.
2 i agree that tests that give the same (say 10% variation) results for 
all languages are bad tests.

however, ''similar results'' is sufficently vague to make me wonder. ie, 
if some languages are twice as fast, but stay on the same curve 
(provided we get real graphs etc), are they similar and the test bad?


bengt