Add Uniform and UniformRM instances.

[vladov3000]

Resolves #182.

[RyanGlScott]

Thanks for the PR!

This compiles with random-1.3.*, but it does not compile with random-1.2.*, as seen in the CI. This is because the uniformRM method of UniformRange didn't receive a Generic-based default until random-1.3.0.

As the ecosystem is still catching up with random-1.3.*, I think it would be worthwhile to provide compatibility with random-1.2.* by explicitly implementing uniformRM in each instance. Would you be willing to do this? I don't think it would be too difficult to do—for instance, I believe this implementation would suffice for V2:

instance UniformRange a => UniformRange (V2 a) where
  uniformRM (V2 x1 y1, V2 x2 y2) g = V2 <$> uniformRM (x1, x2) g <*> uniformRM (y1, y2) g

---

Also, the Uniform and UniformRange classes were introduced in random-1.2.*, so make sure to bump the lower version bounds on random to >= 1.2 in the .cabal file.

[Taneb]

How hard would it be to also add instances for Linear.V.V?

[vladov3000]

In my new commit, I added explicit uniformRM implementations and bumped up the version of random. I also added an instance of Uniform and UniformRM for Linear.V.V.

[vladov3000]

Also, we need conditional compilation of the isInRange method which was only added in random-1.3.0. I built with both a 1.2 and 1.3 version to check that it works. Why do you set the default value of MIN_VERSION_* macros to 1?

[RyanGlScott]

Thanks! This looks good to me.

Why do you set the default value of MIN_VERSION_* macros to 1?

Judging from fbef54e, I think this was a technique that was required for old versions of HLint and when trying to compile the library code with bare GHC (way back when that was a somewhat common thing to do). Nowadays, HLint can handle CPP just fine, and building library code with a build tool like cabal is pretty much the standard. As such, we could probably remove these MIN_VERSION_* macro redefinitions. (No need to do so as part of this PR, however. I can do that separately.)