Justus Sagemüller

Why the need for anything like SDL? Couldn't we just have something quite like [imageReader](http://hackage.haskell.org/package/ffmpeg-light-0.11.1/docs/Codec-FFmpeg.html#v:imageReader), except that it would load chunks of audio data instead of `Image`s?

Kudos for all that hard work! I think an FFmpeg library should keep as close to what FFmpeg spits out as possible while being still _safe_ to use, and not...

Blind guess: “_bunch of distortion and doubles the length_” sounds like stereo-interlacing gone awry. Does the mp3 file say it's mono?

@nick8325 it's true that QuickCheck has very modest dependencies, but I still wouldn't call it a really minimalistic package. It may be a bit silly to not want to depend...

One application I could see outside testing is _showing examples of how a function behaves_, for automated documentation generation etc..

@Soupstraw that's not really an argument, because there are other dedicated libraries that offer random monads. Check out [random-fu](https://hackage.haskell.org/package/random-fu).

@brandonchinn178 I don't think it's a good idea to use ``arbitrary `suchThat` ...`` for generating random values in general code. This is inefficient and you don't really have any control...

To me, the most sensible fix seems to be just not using `np.array` at all. Product spaces are inhomogenenous, and normally each constituent space is at least somewhat big. This...

Probably, object-arrays are actually the less painful option. ```diff --- a/odl/util/testutils.py +++ b/odl/util/testutils.py @@ -323,7 +331,7 @@ def noise_array(space): """ from odl.space import ProductSpace if isinstance(space, ProductSpace): - return np.array([noise_array(si)...

@ozanoktem in this case the problem isn't so much about the behaviour of `ProductSpace` itself, as about the reference arrays that the test suite uses for checking whether numerical operations...