Mark nub and sort functions as INLINE

[toyboot4e]

Fixes #50. Notes:

• nubByMut is marked as INLINABLE. Change it to INLINE if it's faster in some case.
• sortUniq is marked as INLINE, but without any benchmark. I guess it's ok as it's very small.

[erikd]

Something weird with cabal and tests. I have fixed this on master. Please rebase and force push.

[toyboot4e]

I've done rebase and force push. Btw the 七味唐がらし made me smile ha ha.

[toyboot4e]

Simplifier ticks exhausted

Sounds like too much inlining, as functions called from sort are also inlined?

[erikd]

Btw the 七味唐がらし made me smile ha ha.

Where? I have not even seen it. Was not me.

[toyboot4e]

Ha ha I saw it in your profile picture (👉 🍶).

About this PR,

• The nub family change would be ok.
• The sort family change can be reverted if you don't like it (as the test fails for ghc-8.0.2).
  • I guess they should rather change the internal implementation from INLINE to INLINABLE + stToPrim, however, it would be another benchmark and a PR.

[erikd]

Ha ha I saw it in your profile picture

Ah didn't even know what it was called. I just know I like it on eg Teriaki Chciken.

[erikd]

I will merge this as is and remove the ghc-8.0 CI job.

[erikd]

Version 0.9.1.0 uploaded to Hackage.