[WIP] Add cats.Parallel instance for Future

[felixbr]

Hi,

this is a first attempt at adding a cats.Parallel instance (see #90).

Since I'm mostly familiar with Future and not Rerunnable I implemented it for Future first and if that is done I can maybe try to replicate it for Rerunnable.

I based the implementation on the implementations for cats.effect.IO and monix.eval.Task as I know those libraries reasonably well.

I consider this PR WIP as I have some questions:

1. I reused the Newtype concept present in cats-effect and monix. Both have an Apache2.0 license, so I'm not sure how attribution for this piece of code works. Maybe we can just use it, maybe we need to keep the copyright header, I'm not too familiar how this works, sorry.

2. I put the tests next to the existing ones for Future. As ArbitraryInstances doesn't have FuturePar in scope I couldn't move the instances there. For the moment they are in the FutureSuite, but if you don't like this I can put more time into finding a scoping solution to extract it.

3. The elephant in the room is the fact that ap and product on the normal Future instance are already defined in terms of Future.join, so they run things in parallel already. This means traverse and parTraverse as well as sequence and parSequence will all be parallel. This is in line with the behavior of instances of scala.concurrent.Future iirc.
   There is the option to change traverse to be sequential, but sequence cannot be sequential (afaik), so this would be awkward and besides that it'd also be a breaking change.
   I'd argue that parTraverse and parSequence still have the benefit of explicitly signaling the reader how they will execute, even if they don't necessarily bring new functionality to the table.

Hope this is reasonable. Looking forward to your feedback :)

Cheers
~ Felix

[felixbr]

So travis-ci says Expected feature release number in range of 9 to 14, but got: 8.

I'm not familiar with your workflow here, so if I need to PR on something other than master or increment the release number somewhere, you'll have to tell me.

[travisbrown]

@felixbr I think that's a Travis CI JVM thing—they've been phasing out Oracle JDK 8 and I guess it hadn't hit this repo before now. If you change the Travis config to openjdk8 it should work.

[travisbrown]

(I'll review tomorrow—thanks for doing this!)

[codecov-io]

Codecov Report

Merging #132 into master will increase coverage by 0.1%.
The diff coverage is 100%.

@@            Coverage Diff            @@
##           master     #132     +/-   ##
=========================================
+ Coverage    93.8%   93.91%   +0.1%     
=========================================
  Files           4        5      +1     
  Lines         113      115      +2     
  Branches        3        3             
=========================================
+ Hits          106      108      +2     
  Misses          7        7

Impacted Files	Coverage Δ
...main/scala/io/catbird/util/internal/Newtype1.scala	100% <100%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f975e60...81bd263. Read the comment docs.

[travisbrown]

Sorry, lost track of this…

[travisbrown]

We don't support 2.10, so couldn't these go directly on FuturePar?

[felixbr]

Good point, I'll take a look at it again

[felixbr]

Sorry, lost track of this…

No worries, I thought you were busy with your new job 😄

[travisbrown]

Merging this now—we can look at the newtype encoding in a follow-up.