Updates traversable to test the correct laws

[solomon-b]

Resolves issue #24

I wasn't able to implement the Naturality test correctly, but I left it in as a comment in case someone else can get it working. I think it follows from Identity and Composition so we are okay without it.

[sjakobi]

Thanks for your PR! :)

I think it follows from Identity and Composition so we are okay without it.

Do you know a proof?

I'm fine with scrapping support for GHC 7.10 BTW, so you can use TypeApplications.

[sjakobi]

BTW, it would be nice to have tests for the sequenceA laws, too. Feel free to add them, but there's no obligation. :)

[solomon-b]

I think it follows from Identity and Composition so we are okay without it.

Do you know a proof?

hmm, i actually haven't been able to find a proof, but this link claims it is guaranteed to hold:
https://en.wikibooks.org/wiki/Haskell/Traversable#The_Traversable_laws

[sjakobi]

Thanks!

One more nit…

[sjakobi]

Thanks! :)

Since this is a breaking change, I'll merge this after cutting the v0.5.7 release (#64).

[solomon-b]

Great!

[sjakobi]

Regarding naturality, note this change: https://gitlab.haskell.org/ghc/ghc/-/merge_requests/7492

[solomon-b]

Well there you go. Should we hold off on merging this until a working implementation is produced?

[sjakobi]

Well there you go. Should we hold off on merging this until a working implementation is produced?

When do you expect it to land?

IMHO the existing changes are good additions which shouldn't be held back for very long.

[solomon-b]

Well there you go. Should we hold off on merging this until a working implementation is produced?

When do you expect it to land?

IMHO the existing changes are good additions which shouldn't be held back for very long.

To be honest, I'm a little stumped on this how to fix it. Is there anyone with more quickcheck experience who we can ask? The issue is the rank2 type.

[sjakobi]

Published in v0.6.0: https://hackage.haskell.org/package/checkers-0.6.0