News for hackers. Stuff that matters.
Re:Confession of a Haskell Hacker by JadeNB on Sunday July 08, 2012 @02:13PM (#4215171)
> Until the code has actually been tested, he doesn't know that it works, he just thinks that it does.

After the code has been tested, he still doesn't know that it works, only that it sometimes produces the expected output. (As a mathematician, it's a lot easier for me to trust a proof than an empirical verification; but I agree that what the HM type system proves is not always what one wants it to prove.)

Re:Confession of a Haskell Hacker by scott_s on Sunday July 08, 2012 @03:29PM (#4215388)
I agree. I claim that he knows that it works under the circumstance that he has tested, and can be more confident that it will work in circumstances similar to what he has tested.
Re:Confession of a Haskell Hacker by zopa on Monday July 09, 2012 @08:38AM (#4217819)
True. But with very general code, like this library, it would be difficult for unit tests to cover more than a small fraction of the domain.