Latest stack fails to build on GHC 7.10.2 due to bad version of mono-traversable (transitive dependency)

[tebello-thejane]

Trying to install stack on a clean GHC 7.10.2 (no global packages) tries to build mono-traversable-0.4.0, which fails to build on GHC 7.10.2.

Preprocessing library mono-traversable-0.4.0...
[1 of 6] Compiling Data.MonoTraversable ( src/Data/MonoTraversable.hs, dist/build/Data/MonoTraversable.o )

src/Data/MonoTraversable.hs:39:1: Warning:
    The import of ‘$, replicate’ from module ‘Prelude’ is redundant

src/Data/MonoTraversable.hs:72:1: Warning:
    Module ‘Control.Monad.Trans.Error’ is deprecated:
      Use Control.Monad.Trans.Except instead

src/Data/MonoTraversable.hs:131:24: Warning:
    In the use of type constructor or class ‘ErrorT’
    (imported from Control.Monad.Trans.Error):
    Deprecated: "Use Control.Monad.Trans.Except instead"

src/Data/MonoTraversable.hs:192:36: Warning:
    In the use of type constructor or class ‘ErrorT’
    (imported from Control.Monad.Trans.Error):
    Deprecated: "Use Control.Monad.Trans.Except instead"

src/Data/MonoTraversable.hs:324:32: Warning:
    In the use of ‘Unsafe.inlinePerformIO’
    (imported from Data.ByteString.Internal):
    Deprecated: "If you think you know what you are doing, use 'unsafePerformIO'. If you are sure you know what you are doing, use 'unsafeDupablePerformIO'. If you enjoy sharing an address space with a malevolent agent of chaos, try 'accursedUnutterablePerformIO'."

src/Data/MonoTraversable.hs:326:29: Warning:
    In the use of ‘Unsafe.inlinePerformIO’
    (imported from Data.ByteString.Internal):
    Deprecated: "If you think you know what you are doing, use 'unsafePerformIO'. If you are sure you know what you are doing, use 'unsafeDupablePerformIO'. If you enjoy sharing an address space with a malevolent agent of chaos, try 'accursedUnutterablePerformIO'."
makeCorePair: arity missing $cofoldMap_abvQ
[2 of 6] Compiling Data.GrowingAppend ( src/Data/GrowingAppend.hs, dist/build/Data/GrowingAppend.o )
[3 of 6] Compiling Data.Sequences   ( src/Data/Sequences.hs, dist/build/Data/Sequences.o )

src/Data/Sequences.hs:14:1: Warning:
    The import of ‘not’ from module ‘Prelude’ is redundant
[4 of 6] Compiling Data.MinLen      ( src/Data/MinLen.hs, dist/build/Data/MinLen.o )

src/Data/MinLen.hs:80:1:
    Couldn't match representation of type ‘f1 mono’
                             with that of ‘f1 (MinLen nat mono)’
    arising from trying to show that the representations of
      ‘(Element mono -> f1 (Element mono)) -> mono -> f1 mono’ and
      ‘(Element mono -> f1 (Element mono))
       -> MinLen nat mono -> f1 (MinLen nat mono)’ are the same
    Relevant role signatures:
      type role Element nominal
      type role MinLen phantom representational
    NB: We cannot know what roles the parameters to ‘f1’ have;
      we must assume that the role is nominal
    Relevant bindings include
      otraverse :: (Element (MinLen nat mono)
                    -> f (Element (MinLen nat mono)))
                   -> MinLen nat mono -> f (MinLen nat mono)
        (bound at src/Data/MinLen.hs:80:1)
    In the expression:
        ghc-prim-0.4.0.0:GHC.Prim.coerce
          (otraverse ::
             (Element mono -> f (Element mono)) -> mono -> f mono) ::
          forall (f :: * -> *). GHC.Base.Applicative f =>
          (Element (MinLen nat mono) -> f (Element (MinLen nat mono)))
          -> MinLen nat mono -> f (MinLen nat mono)
    In an equation for ‘otraverse’:
        otraverse
          = ghc-prim-0.4.0.0:GHC.Prim.coerce
              (otraverse ::
                 (Element mono -> f (Element mono)) -> mono -> f mono) ::
              forall (f :: * -> *). GHC.Base.Applicative f =>
              (Element (MinLen nat mono) -> f (Element (MinLen nat mono)))
              -> MinLen nat mono -> f (MinLen nat mono)
    When typechecking the code for  ‘otraverse’
      in a derived instance for ‘MonoTraversable (MinLen nat mono)’:
      To see the code I am typechecking, use -ddump-deriv
    In the instance declaration for ‘MonoTraversable (MinLen nat mono)’

src/Data/MinLen.hs:80:1:
    Couldn't match representation of type ‘m1 mono’
                             with that of ‘m1 (MinLen nat mono)’
    arising from trying to show that the representations of
      ‘(Element mono -> m1 (Element mono)) -> mono -> m1 mono’ and
      ‘(Element mono -> m1 (Element mono))
       -> MinLen nat mono -> m1 (MinLen nat mono)’ are the same
    Relevant role signatures:
      type role Element nominal
      type role MinLen phantom representational
    NB: We cannot know what roles the parameters to ‘m1’ have;
      we must assume that the role is nominal
    Relevant bindings include
      omapM :: (Element (MinLen nat mono)
                -> m (Element (MinLen nat mono)))
               -> MinLen nat mono -> m (MinLen nat mono)
        (bound at src/Data/MinLen.hs:80:1)
    In the expression:
        ghc-prim-0.4.0.0:GHC.Prim.coerce
          (omapM :: (Element mono -> m (Element mono)) -> mono -> m mono) ::
          forall (m :: * -> *). GHC.Base.Monad m =>
          (Element (MinLen nat mono) -> m (Element (MinLen nat mono)))
          -> MinLen nat mono -> m (MinLen nat mono)
    In an equation for ‘omapM’:
        omapM
          = ghc-prim-0.4.0.0:GHC.Prim.coerce
              (omapM :: (Element mono -> m (Element mono)) -> mono -> m mono) ::
              forall (m :: * -> *). GHC.Base.Monad m =>
              (Element (MinLen nat mono) -> m (Element (MinLen nat mono)))
              -> MinLen nat mono -> m (MinLen nat mono)
    When typechecking the code for  ‘omapM’
      in a derived instance for ‘MonoTraversable (MinLen nat mono)’:
      To see the code I am typechecking, use -ddump-deriv
    In the instance declaration for ‘MonoTraversable (MinLen nat mono)’
cabal: Error: some packages failed to install:
chunked-data-0.2.0 depends on mono-traversable-0.4.0 which failed to install.
conduit-combinators-1.0.2 depends on mono-traversable-0.4.0 which failed to
install.
mono-traversable-0.4.0 failed during the building phase. The exception was:
ExitFailure 1
stack-0.1.3.0 depends on mono-traversable-0.4.0 which failed to install.

[snoyberg]

This should really be a bug report against cabal, not stack. There's a bug in the cabal dependency solver leading it to make this selection. As usual: I recommend building stuff with stack, not cabal, to avoid dependency hell ;).

[tebello-thejane]

Hehe.

So what's the cabal bug causing this? How is cabal not behaving as expected?

And how do I install a global application using stack, outside of a stack-ified project?

[snoyberg]

I don't know what the cabal bug is, but it's certainly problematic that it's deciding to use a more-than-year-old version of mono-traversable when more recent versions work just fine. You can probably work around this with cabal install --constraint 'mono-traversable >= 0.9' stack.

There's a guide to upgrading stack at https://github.com/commercialhaskell/stack/wiki/Downloads#upgrade. Does stack upgrade not work for you?

[tebello-thejane]

Well, of course stack install stack works outside of a stack-ified project.

Would be nice to have stack build via cabal, though. We haven't abandoned cabal yet.

[tebello-thejane]

stack install stack fails with

Using latest snapshot resolver: lts-2.22
Writing global (non-project-specific) config file to: /home/thejanet/.stack/global/stack.yaml
Note: You can change the snapshot via the resolver field there.
Downloaded lts-2.22 build plan.    
Caching build plan
While constructing the BuildPlan the following exceptions were encountered:

--  While attempting to add dependency,
    Could not find package path in known packages

--  Failure when adding dependencies:    
      base: needed (<0 && <0), latest is 4.8.1.0, but 4.7.0.2 found
      path: needed (-any && >=0.5.1), latest is 0.5.2, but not present in build plan
    needed for package: stack-9.9.9

Recommended action: try adding the following to your extra-deps in /home/thejanet/.stack/global/stack.yaml
- path-0.5.2

You may also want to try the 'stack solver' command

which I guess makes sense.

But stack upgrade seems to works just fine (it's building as I type this with no hitches so far). Thanks. :-)

[snoyberg]

I'm in favor in general of getting stack to build with cabal (that's why the Travis build uses cabal). But I'm not going to spend a lot of effort on it, especially when there's a bug in cabal. If someone else wants to look into this, I'd be greatful.

[tebello-thejane]

I'll try to whip up a bug report for cabal. Thanks.

[tebello-thejane]

Reported here: haskell/cabal#2759

Could we at least note this somewhere, so that new users won't find themselves unable to install stack for the first time? I would not have been able to install the latest stack had I not already had stack installed.

[borsboom]

I've added a note to the README. Note that we generally recommend new users use the binaries rather than building from source (which requires a pre-existing GHC and cabal-install).