[Shootout-list] regexmatch ghc again
Bengt Kleberg
bengt.kleberg@ericsson.com
Mon, 23 May 2005 10:25:58 +0200
On 2005-05-21 03:21, Jon Harrop wrote:
...deleted
> Pidigits doesn't cause problems at the moment. Hopefully future compilers will
> be effective enough to optimise away the pidigits benchmark completely. Until
> then, I think it would be wise to design the benchmarks so that this will
> never be possible.
it is difficult to be future compatible. if possible to design such a
test (without getting bogged down in very details that are not part of
what one wants to measure), then this is good.
but i am reluctant to demand that all test must be safe for future
possiblities. i would not be able to do it.
bengt