With such a big speedup and not a lot of downside, I'd really rather it wind up central enough to be used in preference to nub wherever possible (including in base!). Perhaps they can duplicate the minimum functionality necessary from Data.Set, internally behind the scenes? Alternatively, maybe we can move nub out of base?

On Fri, Oct 18, 2013 at 11:47 AM, Yitzchak Gale <gale@sefer.org> wrote:

Niklas Hambüchen wrote:
> It would need a set in base. I wonder if it was ok to use the Data.Set
> implementation internally in base, without exposing it.

Ah yes, this is what stopped previous attempts at getting
ordNub in.

The GHC team is very keen on getting more things *out* of base.
Unless something has changed, getting them to include new things
in base, even if hidden, would be extremely difficult.
Adding additional dependencies and maintenance burdens would
add extra work for the overburdened GHC team and slow the
GHC release cycle.

How about adding a module named something like Data.List.Nub
to containers and putting ordNub in there? Or just go back to the
idea of putting it in Data.Set.

-Yitz

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe