Various changes to make it compile with MicroHs.

[augustss]

Mostly #ifdefs and some type signatures.

[treeowl]

Thank you! This is a very welcome development. I don't see any changes to CI. Could we at least compile the library on MicroHS in CI, even if dependencies don't yet let us compile the test suite?

[augustss]

Hmm. That's a good idea. I will have to look into how easy it is. Can CI compile other Hackage packages?

…

On Thu, Sep 19, 2024 at 2:06 PM David Feuer ***@***.***> wrote: Thank you! This is a very welcome development. I don't see any changes to CI. Could we at least compile the library on MicroHS in CI, even if dependencies don't yet let us compile the test suite? — Reply to this email directly, view it on GitHub <#1043 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABIARHUNDCBYRVTVWYG6S53ZXK45NAVCNFSM6AAAAABOPRDUIGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNRQHAYDSOJSGA> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[treeowl]

I imagine so, though I'm not at all an expert in the ways of haskell-ci.

[meooow25]

What exactly are the differences between getting something to compile on GHC and MicroHs. It seems to be mostly about certain modules not available in MicroHs base?

And +1 on having a CI so we don't break this in the future.

[meooow25]

By the way, I am having to manually approve these CI runs. It seems that this is the default for "first time contributors" and I don't see a way to change it. treeowl could you peek at the settings to see if we can change this (perhaps temporarily) to make things easier?

[treeowl]

I don't think I have the authority to change that. I think we'd need a haskell owner or something.

[augustss]

Yes, getting it to run with MHS is mostly a matter of #ifdef-ing out things that base does not support (and cannot support right now, since it needs extensions like QualifiedConstraints).
There's also a semantics difference to the scoping of type variables in signatures. I might change that for easier porting.

[treeowl]

Would it make sense on MicroHs to have a local copyish of Control.Category to limit the CPP and offer that interface more universally?

[augustss]

I've added CI for MHS. It's a little clunky right now, but I hope to improve it.

[augustss]

I'll add Control.Category and update the PR.

[meooow25]

If QuantifiedConstraints is a problem, MicroHs could define the pre-QuantifiedConstraints versions of the classes (Eq1, Ord1, Bifoldable, etc). This would let us get rid of most of the CPP.

[augustss]

Just for your amusement: containers has 25kloc of Haskell code and MHS has 11kloc. :)

[augustss]

I'm making the MHS base library more GHC compatible. The PR will be updated soon.

[treeowl]

Don't compromise your principles too much, but the less CPP we have to deal with the better.

[augustss]

I've made my changes. Please have a look again.

[meooow25]

Thanks for the updates, this is looking much nicer!