Support GHC 9.2.

[patrickt]

Widens template-haskell bounds and removes mappend definitions in favor of defaulting to <>, as per -Wnoncanonical-monoid-instances.

[patrickt]

This is ready for review.

[patrickt]

@jacobstanley ping

[weebl2000]

Cool.

[tmcgilchrist]

Could you add this GHC version to the GH CI setup so it gets fully tested.
https://github.com/hedgehogqa/haskell-hedgehog/blob/master/.github/workflows/ci.yaml#L17

Seems like a maintainer needs to approve the PR so CI gets kicked off.

[patrickt]

@jacobstanley Any word on this?

[smunix]

@moodmosaic ping

[moodmosaic]

👍 🚀 🍻

[moodmosaic]

It looks like some builds are failing. Have to investigate on that.

[gwils]

It looks as though you will need to use CPP to give mappend an explicit definition on GHC 8.2 and 8.0, in order to make -Werror happy.

[moodmosaic]

It looks as though you will need to use CPP to give mappend an explicit definition on GHC 8.2 and 8.0, in order to make -Werror happy.

Could you do that, @patrickt? Otherwise we'll get to it as soon as possible.

[patrickt]

@moodmosaic On it!

[ysangkok]

@patrickt It is better to define mappend = (<>) on versions of base where Semigroup is a superclass of Monoid. Otherwise this warning is triggered: https://downloads.haskell.org/~ghc/9.2.1/docs/html/users_guide/using-warnings.html#ghc-flag--Wnoncanonical-monoid-instances . I believe this would be GHC 8.0 and 8.2 according to the linked docs. EDIT: The last sentence was meant to refer to versions where it is not a superclass.

[patrickt]

@moodmosaic I’ve tested this with 8.2 and 8.0 locally; can I get another approval/merge if it looks good to you?

[moodmosaic]

@moodmosaic I’ve tested this with 8.2 and 8.0 locally; can I get another approval/merge if it looks good to you?

Sure, I just run it. 🍻

[moodmosaic]

@patrickt, some are still failing but we're almost there.

[patrickt]

@moodmosaic Ugh, my bad. This last one should be okay.

[moodmosaic]

@moodmosaic Ugh, my bad. This last one should be okay.

No problem! Hmm, I think there's one more left to go 👀

[moodmosaic]

@patrickt, I know this is getting boring at this point 😴 It looks like some checks are still failing.

[moodmosaic]

Yep, looking at the build error it looks like changes in 94dd5b8 must also be applied in Registry.hs. @patrickt, let me know if you'd want me to do that.

[ysangkok]

@patrickt looks like a new warning in 9.2 makes the build fail:

src\Test\Example\Resource.hs:174:7: error: [-Wincomplete-uni-patterns, -Werror=incomplete-uni-patterns]
Error:     Pattern match(es) are non-exhaustive
    In a pattern binding:
        Patterns of type ‘([String], [String])’ not matched:
            ([], [])
            ((_:_), [])
    |
174 |       (xs, x : ys) =

    |       ^^^^^^^^^^^^^^^...

[patrickt]

Sorry for the churn, everyone. I’ve turned on -Werror locally and observed a clean build. Hopefully we’re good to go.

[gwils]

Perhaps a maintainer could trigger a CI build to verify this? @moodmosaic

[moodmosaic]

Sorry for the churn, everyone. I’ve turned on -Werror locally and observed a clean build. Hopefully we’re good to go.

No problem at all! 🍻

[moodmosaic]

Perhaps a maintainer could trigger a CI build to verify this? @moodmosaic

@gwils, CI build started 🚀

[moodmosaic]

Thank you, @patrickt 🍻 Also thank you @ysangkok, @gwils, @tmcgilchrist for the great feedback 🍺

[ysangkok]

I think this was released as part of https://github.com/hedgehogqa/haskell-hedgehog/releases/tag/1.1, with bounds similar to the ones in this PR at https://hackage.haskell.org/package/hedgehog