Proposal: Allow gunfold for Data.Map, Data.IntMap, etc.

Edward Kmett ekmett at gmail.com
Sat Sep 1 14:23:07 CEST 2012


For what it is worth, Milan already implemented the smarter fromList and it
is considerably faster. =)

If I'm reading his numbers right, its ~50% faster than the original in the
worst case, and its about ~10x faster on large sorted inputs, and within 5%
of the speed of fromDistinctAscList.

Overall, it is a huge win.

-Edward

On Thu, Aug 30, 2012 at 9:50 AM, Brent Yorgey <byorgey at seas.upenn.edu>wrote:

> On Wed, Aug 29, 2012 at 06:06:25PM -0700, Johan Tibell wrote:
> > On Wed, Aug 29, 2012 at 5:57 PM, Edward Kmett <ekmett at gmail.com> wrote:
> > > This might pay off as well, but I am leery that its a rather tricky
> > > balancing act and will take a lot of profiling to find the right
> balance in
> > > practical performance vs. asymptotic bounds.
> > >
> > > Should we split the notion of improving the performance of fromList
> into a
> > > separate project/proposal
> > > that simply exposes synergies with this one?
> >
> > Sure. Lets continue this thread on its original topic. I'm +1 on the
> > proposal.
>
> +1 from me too.  A smarter fromList also sounds cool but I agree that
> should be a separate proposal, if only to avoid derailing this one.
>
> -Brent
>
> _______________________________________________
> Libraries mailing list
> Libraries at haskell.org
> http://www.haskell.org/mailman/listinfo/libraries
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.haskell.org/pipermail/libraries/attachments/20120901/66b2f0a5/attachment.htm>


More information about the Libraries mailing list