Open alexanderGugel opened 9 years ago
Hmm... thanks for pointing this out @alexanderGugel, I should probably fix this, though I can't possibly see how it could fail the test. This might be an issue with Vec3.normalize I'll look into it.
@redwoodfavorite Hmmm... I also tried to reproduce the failing test locally, but couldn't get it to fail. This seems very strange to me. In either case, testing using randomized values is fine IMO, but we should log them.
@redwoodfavorite are you actively working on this?
Currently there are a couple of non-deterministic tests that should either be made deterministic, fixed or adjusted to log the generated random values.
Example
in
GeometryHelper.spec.js
The issue is
var vertices = generateRandomArray(10, [-10, 10])
. Sometimes it returns an array that makes the test fail, sometimes it doesn't (in which case we don't know about the values that made it fail). I've seen this before, but wasn't sure it was actually related to this test, but it seems to be the case. E.g. I'm specifically talking about the following CI build that passes on 0.10, but fails on iojs: https://travis-ci.org/Famous/engine/builds/68386307@michaelobriena How do you want to proceed with this?