[Haskell-cafe] [GSOC] About "Implementing Maps using generalized tries" project idea

[Jamie Brandon]

http://hackage.haskell.org/cgi-bin/hackage-scripts/package/gmap-0.1

Everything in the release is working. There's some fast map/trie
implementations and a bunch of map combinators. In Test.GMap there are
~800 lines of quickcheck properties. The tests use some undocumented
type hackery to allow them to run on any instance of the GMap class.
Some of the strictness invariants demanded by the test suite are not
documented.

There are no real benchmarks. I'm still not even sure how to
effectively benchmark haskell code.

If you want to suggest this as a project for 2009 I think the
priorities should be:
  Documentation + examples (which I'm willing to help with if there is interest)
  Systematic benchmarks
  Integration with the edison api, to encourage a single collection
api on hackage

Jamie