Hacker News Reader

Maybe there's a place for some sort of a smart test runner that would execute assertions multiple times with random input, trying to achieve failure. If all attempts succeed then it'd mark test as questionable.

View on HN

Have you heard of fuzzing?[1]

1. http://en.wikipedia.org/wiki/Fuzz_testing

I haven't, thanks for pointing out. Seems like a closely related concept, although appears to be used mostly in security testing context. Also, at least for Python most tools are a bit too complex—I imagined something more like a TestCase subclass (not even a test runner) that augments certain assert methods and possibly provides new ones. Anyway, on a second thought, adopting TDD looks like a better solution.

There's this: http://ruby.sadi.st/Heckle.html

It changes your class, then runs your tests to make sure they fail. I haven't tried it myself.

Reply on news.ycombinator.com