gale at sefer.org
Thu Oct 17 16:56:39 UTC 2013
Niklas Hambüchen wrote:
> I make the following, really cool proposal:>
> * You add your test cases that you just mentioned to the benchmark list
> in "ordnub.hs"...
No thanks. Those were not meant to be test cases to add
to a benchmark suite - they were meant to get you thinking a
little more deeply about execution models of Haskell in
general, and the GHC execution model in particular, given
your claim that nubOrd is better than ord is *every* case,
in *every* sense of "better".
But that is a side issue. Everyone agrees with you
that it is useful to add a good implementation of nubOrd to
the standard libraries in some convenient and logical
place. Let's get back to focusing on that.
More information about the Haskell-Cafe