hls-fourmolu-plugin-1.1.1.0 is incompatible with ghcide-1.8.0.0

[lehins]

Building hls-fourmolu-plugin-1.1.1.0 results in a build error, when it is built with haskell-language-server-1.8.0.0:

Preprocessing library for hls-fourmolu-plugin-1.1.1.0..
Building library for hls-fourmolu-plugin-1.1.1.0..
[1 of 2] Compiling Ide.Plugin.Fourmolu.Shim ( src/Ide/Plugin/Fourmolu/Shim.hs, dist/build/Ide/Plugin/Fourmolu/Shim.o, dist/build/Ide/Plugin/Fourmolu/Shim.dyn_o )
[2 of 2] Compiling Ide.Plugin.Fourmolu ( src/Ide/Plugin/Fourmolu.hs, dist/build/Ide/Plugin/Fourmolu.o, dist/build/Ide/Plugin/Fourmolu.dyn_o )

src/Ide/Plugin/Fourmolu.hs:62:56: error:
    Variable not in scope:
      usePropertyAction
        :: t0
           -> PluginId
           -> Properties '[ 'PropertyKey "external" 'TBoolean]
           -> Action Bool
   |
62 |     useCLI <- liftIO $ runAction "Fourmolu" ideState $ usePropertyAction #external plId properties
   |                                                        ^^^^^^^^^^^^^^^^^

Which makes sense, since usePropertyAction was not exported until 1.8.0.2, which is not on hackage:

haskell-language-server/ghcide/src/Development/IDE.hs

Lines 19 to 20 in ddb21a0

	import Development.IDE.Core.Rules as X (getClientConfigAction,
	getParsedModule)

I'd recommend creating a revision on hackage that fixes the lower bound: ghcide >= 1.9

[georgefst]

Good catch.

I'd recommend creating a revision on hackage that fixes the lower bound: ghcide >= 1.9

Any reason not to make it ghcide >= 1.8.0.2? It's more correct, and a user could always supply 1.8.0.2 locally.

[lehins]

Any reason not to make it ghcide >= 1.8.0.2?

Technically semantic changes should not be included with last component bump according to PVP, therefor such release should have been 1.8.1.0, not 1.8.0.2. However, since it was never officially released on hackage it is hard to argue about it

[michaelpj]

Fixed