chore(nix): Bump to haskell.nix experimental GHC 9.6 branch.

[dhess]

No description provided.

[brprice]

I pushed a commit fixing logging-effect (using a unmerged upstream PR). We now seem to be hitting jfischoff/postgres-options#6 (which is odd -- it seems that the code on hackage is newer than that on github?)

[brprice]

There is also an interesting new warning in primer, which I minimized to: the following code warns:

{-# LANGUAGE DerivingStrategies #-}
{-# LANGUAGE GeneralizedNewtypeDeriving #-}
{-# OPTIONS -Wredundant-constraints #-}
module Primer.Database where

import Prelude
import Control.Monad.Reader
import Control.Applicative

newtype NullDbT m a = NullDbT {unNullDbT :: ReaderT Int m a}
  deriving newtype
    ( Functor
    , Applicative
    , Alternative
    , Monad
    , MonadTrans
    )

Bug.hs:16:7-16: warning: [GHC-30606] [-Wredundant-constraints]
    • Redundant constraint: Monad m
    • When deriving the instance for (MonadTrans NullDbT)
   |
16 |     , MonadTrans
   |       ^^^^^^^^^^

I have not yet worked out the issue This looks like a GHC bug to me. I have opened https://gitlab.haskell.org/ghc/ghc/-/issues/23143 upstream (this is fixed in ghc master, but not backported to 9.6.x yet).

[brprice]

I pushed a commit fixing logging-effect (using a unmerged upstream PR).

This is now fixed upstream, so when haskell.nix picks it up, we can replace that commit with one bumping logging-effect to 1.4.0

[dhess]

Closing this as it's out of date: most of these workarounds are no longer necessary.