For code designed not to do anything, HNOP has high hopes. I have recently split the project into two Cabal packages: "nop", a library of no-op services, and "hnop", a program that uses nop to do nothing. Both packages can be found in this repository: darcs get http://semantic.org/hnop/ First and foremost the two packages are intended to be templates for Cabal projects, so I'm interested in making them as canonical and "best practices" as possible. I'm not so interested in showing off clever Haskell programming idioms for doing nothing, unless there's some expository benefit relating to the Cabal or Haddock infrastructure. For instance, in the nop package, "runhaskell Setup test" works. It runs "make" in the "test" directory, which compiles a program and runs it, and then diffs it against reference output. But should I be using QuickCheck or something here? Is there some well-known or Cabal-approved way to run tests on a package? Please send me patches licensed under the simple permissive license in the LICENSE files if you have any improvements. Or I can move this to darcs.haskell.org if it's appropriate. -- Ashley Yakeley Seattle, WA