I'd also like to point to a couple of CSV libraries I released a long time ago and have been maintaining that both target constant-space operation and try (and hope) for the best in terms of speed. I'd be very interested to know how they fare in terms of performance benchmarking:

Latest, based on conduit: http://hackage.haskell.org/package/csv-conduit (just released the latest version)

Older, based on enumerator: http://hackage.haskell.org/package/csv-enumerator

Notice how both are based on IO streaming libraries of fame to achieve both constant space operation AND nice interoperability with their habitat. I have found this to be especially true in the case of conduit.

If you end up designing a benchmark, I'd be happy to get it working with my library.

- Oz

On Monday, February 25, 2013 at 5:16 PM, John Wiegley wrote:

Malcolm Wallace <malcolm.wallace@me.com> writes:

Simple answer - I have never heard of cassava, and suspect it did not exist
when I first did the benchmarking. I'd be happy to re-do my performance
comparison, including cassava and any other recent-ish CSV libraries, if I
can find them.

I would be very interested in those results, Malcolm.

Thanks,
--
John Wiegley
FP Complete Haskell tools, training and consulting
http://fpcomplete.com johnw on #haskell/irc.freenode.net

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