November 2016 - ghc-tickets

[GHC] #8827: Inferring Safe mode with GeneralizedNewtypeDeriving is wrong
by GHC 01 Aug '17

01 Aug '17

#8827: Inferring Safe mode with GeneralizedNewtypeDeriving is wrong ------------------------------------+------------------------------------- Reporter: goldfire | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.8.1 Component: Compiler | Version: 7.8.1-rc1 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: None/Unknown Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | ------------------------------------+------------------------------------- Consider the following modules: {{{ module A (List, ints) where data List a = Nil | Cons a (List a) infixr 4 `Cons` ints :: List Int ints = 1 `Cons` 2 `Cons` 3 `Cons` Nil }}} {{{ {-# LANGUAGE GeneralizedNewtypeDeriving #-} module B where import A newtype Age = MkAge Int deriving C class C a where cList :: List a instance C Int where cList = ints }}} {{{ {-# LANGUAGE Safe #-} module C (ages) where import A import B ages :: List Age ages = cList }}} Module C compiles without a hiccup. But, it shouldn't: the coercion between `ages` and `ints` (performed by !GeneralizedNewtypeDeriving in module B) isn't Safe, as it breaks through `List`'s abstraction. (Note that the constructors of `List` are ''not'' exported from module A!) If module B includes `{-# LANGUAGE Safe #-}`, it duly doesn't compile, because of the stringent "constructors-must-be-in-scope" check done by the `Coercible` mechanism. The problem is that safety can be ''inferred'' without this check taking place. You may also want to read the commentary on #8745 for related discussion. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8827> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 52

[GHC] #12639: Inconsistent treatment of FlexibleInstances and MPTCs with standard vs. flexible deriving
by GHC 01 Aug '17

01 Aug '17

#12639: Inconsistent treatment of FlexibleInstances and MPTCs with standard vs. flexible deriving -------------------------------------+------------------------------------- Reporter: dfeuer | Owner: Type: feature | Status: new request | Priority: normal | Milestone: Component: Compiler | Version: 8.1 (Type checker) | Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Given `{-# LANGUAGE GeneralizedNewtypeDeriving #-}`, I can write {{{#!hs import Control.Monad.State.Strict newtype Foo s m a = Foo (StateT s m a) deriving (Functor, Applicative, Monad, MonadState s) }}} However, if I want to use `StandaloneDeriving` to make the `MonadState` instance more explicit, {{{#!hs deriving instance Monad m => MonadState s (Foo s m) }}} I suddenly need to add `FlexibleInstances` and `MultiParamTypeClasses`. In my personal opinion, the most sensible way to handle this is to change two things in two different directions: 1. Allow MPTC instance declarations (but not class declarations) without `MultiParamTypeClasses`. 2. Require `FlexibleInstances` for standard deriving clauses when they would be required for standalone deriving declarations. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12639> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 2

[GHC] #9790: Produce coercion rules for derived Functor instances
by GHC 01 Aug '17

01 Aug '17

#9790: Produce coercion rules for derived Functor instances -------------------------------------+------------------------------------- Reporter: dfeuer | Owner: Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.9 Keywords: coercion | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Runtime Blocked By: | performance bug Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- I don't ''think'' we do this yet, but I think we probably can. If we derive a `Functor` instance, we should, whenever possible and valid, produce some sort of `fmap coerce = coerce` rule for that type. This is done manually for lists in `GHC.Base`, but I imagine it probably applies (much) more generally. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9790> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #10598: DeriveAnyClass and GND don't work well together
by GHC 01 Aug '17

01 Aug '17

#10598: DeriveAnyClass and GND don't work well together -------------------------------------+------------------------------------- Reporter: osa1 | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.11 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Blocked By: Test Case: | Related Tickets: Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- I think we definitely have a bug here, but I'm not sure what it really is. Here's the program: {{{ newtype MyMaybe a = MyMaybe (Maybe a) deriving (Functor, Show) main = print $ MyMaybe $ Just (10 :: Int) }}} I'm using GHC 7.10.1. {{{ ➜ deriveany_bug ghc --make -fforce-recomp Test.hs -XDeriveAnyClass -XGeneralizedNewtypeDeriving [1 of 1] Compiling Main ( Test.hs, Test.o ) Test.hs:2:13: Can't make a derived instance of ‘Functor MyMaybe’ (even with cunning newtype deriving): You need DeriveFunctor to derive an instance for this class Try GeneralizedNewtypeDeriving for GHC's newtype-deriving extension In the newtype declaration for ‘MyMaybe’ }}} Just to try, changing argument order: {{{ ➜ deriveany_bug ghc --make -fforce-recomp Test.hs -XGeneralizedNewtypeDeriving -XDeriveAnyClass [1 of 1] Compiling Main ( Test.hs, Test.o ) Test.hs:2:13: Can't make a derived instance of ‘Functor MyMaybe’ (even with cunning newtype deriving): You need DeriveFunctor to derive an instance for this class Try GeneralizedNewtypeDeriving for GHC's newtype-deriving extension In the newtype declaration for ‘MyMaybe’ }}} It works fine if I remove `DeriveAnyClass`: {{{ ➜ deriveany_bug ghc --make -fforce-recomp Test.hs -XGeneralizedNewtypeDeriving [1 of 1] Compiling Main ( Test.hs, Test.o ) Linking Test ... }}} GHC HEAD is failing in exactly the same way. User manual is saying this in 7.5.6: > In case you try to derive some class on a newtype, and -XGeneralizedNewtypeDeriving is also on, -XDeriveAnyClass takes precedence. But then why is it telling me to enable `GeneralizedNewtypeDeriving` in the error message? Even if I already enabled it? Also, maybe it could try `GND` when `DeriveAnyClass` fails? Because the doc is saying `DeriveAnyClass` has precedence but doesn't specify what happens if it fails. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10598> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 55

[GHC] #10938: DeriveAnyClass + deriving Bifunctor causes GHC panic
by GHC 01 Aug '17

01 Aug '17

#10938: DeriveAnyClass + deriving Bifunctor causes GHC panic -------------------------------------+------------------------------------- Reporter: erantapaa | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.2 Keywords: | Operating System: MacOS X DeriveAnyClass | Architecture: x86_64 | Type of failure: Compile-time (amd64) | crash Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | -------------------------------------+------------------------------------- Attempting the derive Bifunctor with DeriveAnyClass caused this message from GHC: {{{ $ ghc Foo.hs [1 of 1] Compiling Foo ( Foo.hs, Foo.o ) Var/Type length mismatch: [a_an7, b_an8] [] Var/Type length mismatch: [a_an7, b_an8] [] Var/Type length mismatch: [a_an7, b_an8] [] Foo.hs:8:13:ghc: panic! (the 'impossible' happened) (GHC version 7.10.2 for x86_64-apple-darwin): tcTyVarDetails a_an7 Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug }}} The code: {{{#!hs {-# LANGUAGE DeriveAnyClass #-} module Foo where import Data.Bifunctor data Blah a b = A a | B b deriving (Bifunctor) }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10938> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #11509: Incorrect error message in StandaloneDeriving: "The data constructors of <typeclass> are not all in scope"
by GHC 01 Aug '17

01 Aug '17

#11509: Incorrect error message in StandaloneDeriving: "The data constructors of <typeclass> are not all in scope" -------------------------------------+------------------------------------- Reporter: edsko | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1-rc1 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Consider {{{#!hs {-# OPTIONS_GHC -Wall #-} {-# LANGUAGE ConstraintKinds #-} {-# LANGUAGE FlexibleInstances #-} {-# LANGUAGE GADTs #-} {-# LANGUAGE GeneralizedNewtypeDeriving #-} {-# LANGUAGE KindSignatures #-} {-# LANGUAGE ScopedTypeVariables #-} {-# LANGUAGE StandaloneDeriving #-} {-# LANGUAGE StaticPointers #-} {-# LANGUAGE UndecidableInstances #-} {-# LANGUAGE UndecidableSuperClasses #-} module Bug where import Data.Kind import Data.Typeable import GHC.StaticPtr {------------------------------------------------------------------------------- Standard Cloud-Haskell-like infrastructure See <https://ghc.haskell.org/trac/ghc/wiki/TypeableT> for a dicussion of 'SC'. -------------------------------------------------------------------------------} class Serializable a -- empty class, just for demonstration purposes instance Serializable a => Serializable [a] data Static :: * -> * where StaticPtr :: StaticPtr a -> Static a StaticApp :: Static (a -> b) -> Static a -> Static b staticApp :: StaticPtr (a -> b) -> Static a -> Static b staticApp = StaticApp . StaticPtr data Dict :: Constraint -> * where Dict :: c => Dict c class c => SC c where dict :: Static (Dict c) instance (Typeable a, SC (Serializable a)) => SC (Serializable [a]) where dict = aux `staticApp` dict where aux :: StaticPtr (Dict (Serializable a) -> Dict (Serializable [a])) aux = static (\Dict -> Dict) {------------------------------------------------------------------------------- Demonstrate the bug -------------------------------------------------------------------------------} newtype MyList a = MyList [a] deriving instance (Typeable a, SC (Serializable a)) => SC (Serializable (MyList a)) }}} This gives the following type error: {{{ Bug1.hs:40:1: error: • Can't make a derived instance of ‘SC (Serializable (MyList a))’: The data constructors of ‘Serializable’ are not all in scope so you cannot derive an instance for it • In the stand-alone deriving instance for ‘SC (Serializable a) => SC (Serializable (MyList a))’ }}} This of course doesn't make much sense: `Serializable` is a type class, not a datatype, and doesn't have data constructors. I wasn't sure if this deriving clause was going to work at all, or whether I would expect it to. Since `MyList` is a newtype wrapper around `[a]`, and we have the requisite instance {{{#!hs instance (Typeable a, SC (Serializable a)) => SC (Serializable [a]) }}} I was kind of hoping that `GeneralizedNewtypeDeriving` would work its magic. However, even if it cannot, the error message should probably change. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11509> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 16

[GHC] #12814: Should GND infer an instance context when deriving method-free classes?
by GHC 01 Aug '17

01 Aug '17

#12814: Should GND infer an instance context when deriving method-free classes? -------------------------------------+------------------------------------- Reporter: RyanGlScott | Owner: Type: bug | Status: new Priority: normal | Milestone: 8.2.1 Component: Compiler | Version: 8.0.1 (Type checker) | Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: #11369, #12810 Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- This is a design question that emerged from code that I originally discovered [https://ghc.haskell.org/trac/ghc/ticket/11369#comment:17 here] and [https://ghc.haskell.org/trac/ghc/ticket/12810#comment:3 here]. To recap, it's now possible to have code like this: {{{#!hs {-# LANGUAGE GeneralizedNewtypeDeriving, TypeFamilies #-} class C a where type T a newtype Identity a = Identity a deriving C }}} Compiling this (with `-Wredundant-constraints`) generates this code: {{{#!hs instance C a => C (Identity a) where type T (Identity a) = T a }}} But now GHC will complain: {{{ • Redundant constraint: C a • In the instance declaration for ‘C (Identity a)’ }}} This warning makes sense from the perspective that the `C a` constraint isn't ever used by the associated type family instance. So the question arises: should GND avoid generating an instance context for the representation type in the event it's deriving a class with no methods? Fortunately, patching GHC to do this is trivial: {{{#!diff diff --git a/compiler/typecheck/TcDeriv.hs b/compiler/typecheck/TcDeriv.hs index 4722f16..df2d3d5 100644 --- a/compiler/typecheck/TcDeriv.hs +++ b/compiler/typecheck/TcDeriv.hs @@ -1272,7 +1272,8 @@ mkNewTypeEqn dflags overlap_mode tvs [ let (Pair t1 t2) = mkCoerceClassMethEqn cls dfun_tvs inst_tys rep_inst_ty m in mkPredOrigin (DerivOriginCoerce meth t1 t2) TypeLevel (mkReprPrimEqPred t1 t2) - | meth <- classMethods cls ] + | meth <- meths ] + meths = classMethods cls -- If there are no tyvars, there's no need -- to abstract over the dictionaries we need @@ -1281,7 +1282,11 @@ mkNewTypeEqn dflags overlap_mode tvs -- instance C T -- rather than -- instance C Int => C T - all_preds = rep_pred_o : coercible_constraints ++ sc_theta -- NB: rep_pred comes + all_preds = if null meths then [] else [rep_pred_o] + -- See Note [GND and method-free classes] + ++ coercible_constraints + ++ sc_theta + -- NB: rep_pred_o comes first ------------------------------------------------------------------- -- Figuring out whether we can only do this newtype-deriving thing }}} After implementing this patch, I ran the testsuite, and there were some surprising results. One thing that shocked me was that the program reported in #6088, which had previously failed due to a patch resulting from #8984, was now passing! {{{#!hs {-# LANGUAGE TypeFamilies #-} {-# LANGUAGE GeneralizedNewtypeDeriving #-} {-# LANGUAGE EmptyDataDecls #-} module T6088 where class C a newtype A n = A Int type family Pos n data True instance (Pos n ~ True) => C (A n) newtype B n = B (A n) deriving (C) }}} That is because previously, GHC was trying to generate an instance like this: {{{#!hs instance (Pos n ~ True) => C (B n) }}} And this was rejected, since we don't infer exotic equality constraints when deriving. But with this patch, it now generates: {{{#!hs instance {- Empty context => -} C (B n) }}} Which is certainly valid. But is it what a user would expect? I'm not so sure. As hvr notes in #11369, sometimes empty classes are used to enforce invariants, like in the following case: {{{#!hs class Throws e readFoo :: Throws IOError => FilePath -> IO Foo readFoo fn = {- ... -} }}} What if you wanted to have a `Throws` instance for a newtype? You'd probably want something like this: {{{#!hs newtype Id a = MkId a instance Throws a => Throws (Id a) }}} Which feels like something GND should be able to take care of with ease. But to your surprise, you try this: {{{#!hs newtype Id a = MkId a deriving Throws }}} And now GHC generates not the instance above, but rather just: {{{#!hs instance Throws (Identity a) }}} So it's possible that we would lose some of the expressiveness of GND by implementing this change. Is that acceptable? I'm not sure, so I though I'd solicit feedback here. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12814> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 17

[GHC] #8631: Need ImpredicativeTypes for GeneralizedNewtypeDeriving?
by GHC 01 Aug '17

01 Aug '17

#8631: Need ImpredicativeTypes for GeneralizedNewtypeDeriving? ------------------------------------+------------------------------------- Reporter: goldfire | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.7 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: None/Unknown Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | ------------------------------------+------------------------------------- Nathan Howell posts this code: {{{ {-# LANGUAGE FlexibleInstances #-} {-# LANGUAGE GeneralizedNewtypeDeriving #-} {-# LANGUAGE MultiParamTypeClasses #-} {-# LANGUAGE RankNTypes #-} -- Uncomment to compile on GHC 7.8 -- {-# LANGUAGE ImpredicativeTypes #-} module Repro where import Control.Monad.Trans.Cont import Control.Monad.Trans.State.Lazy newtype AnyContT m a = AnyContT { unAnyContT :: forall r . ContT r m a } class MonadAnyCont b m where anyContToM :: (forall r . (a -> b r) -> b r) -> m a instance MonadAnyCont b (AnyContT m) where anyContToM _ = error "foo" data DecodeState = DecodeState newtype DecodeAST a = DecodeAST { unDecodeAST :: AnyContT (StateT DecodeState IO) a } deriving (MonadAnyCont IO) }}} Compiling on HEAD produces {{{ [1 of 1] Compiling Repro ( repro.hs, interpreted ) repro.hs:24:13: Cannot instantiate unification variable ‛b0’ with a type involving foralls: (forall r. (a1 -> IO r) -> IO r) -> DecodeAST a1 Perhaps you want ImpredicativeTypes In the expression: GHC.Prim.coerce (anyContToM :: (forall (r :: *). (a -> IO r) -> IO r) -> AnyContT (StateT DecodeState IO) a) :: forall (a :: *). (forall (r :: *). (a -> IO r) -> IO r) -> DecodeAST a In an equation for ‛anyContToM’: anyContToM = GHC.Prim.coerce (anyContToM :: (forall (r :: *). (a -> IO r) -> IO r) -> AnyContT (StateT DecodeState IO) a) :: forall (a :: *). (forall (r :: *). (a -> IO r) -> IO r) -> DecodeAST a Failed, modules loaded: none. }}} Two questions: 1. Should we require users to specify !ImpredicativeTypes here? Or, should the !GeneralizedNewtypeDeriving mechanism (which sometimes is impredicative) just assume the extension? 2. Can we improve the error message? I'd like to note that the original example really ''does'' require impredicativity -- the question is whether and how to bother the user with this technicality. Small program note: I've done a lot of stuff with !GeneralizedNewtypeDeriving lately but am on holiday until Jan. 6, so don't expect responses! -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8631> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 17

[GHC] #9444: -ddump-deriv doesn't dump failed newtype-deriving
by GHC 01 Aug '17

01 Aug '17

#9444: -ddump-deriv doesn't dump failed newtype-deriving -------------------------------------+------------------------------------- Reporter: goldfire | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.8.2 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Blocked By: | None/Unknown Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- When I say {{{#!hs {-# LANGUAGE GeneralizedNewtypeDeriving, TypeFamilies #-} {-# OPTIONS_GHC -ddump-deriv #-} type family F a type instance F Int = Bool class C a where meth :: a -> F a instance C Int where meth = (> 0) newtype Age = MkAge Int deriving C }}} I get an error (which I should), but I don't get the derived `C` instance, as requested by `-ddump-deriv`. The output should include this helpful information. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9444> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #10361: DeriveAnyClass does not fill in associated type defaults
by GHC 01 Aug '17

01 Aug '17

#10361: DeriveAnyClass does not fill in associated type defaults -------------------------------------+------------------------------------- Reporter: kosmikus | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.1 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Blocked By: Test Case: | Related Tickets: Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- I would expect `test1` and `test2` below to typecheck. This is a reduced test case from trying to use `DeriveAnyClass` on the `Generic` class of the `generics-sop` package, which unfortunately fails due to this bug. {{{ {-# LANGUAGE DeriveAnyClass, StandaloneDeriving, TypeFamilies #-} module Test where class C1 a where type T1 a type instance T1 a = Char class C2 a where -- equivalent to C1 type T2 a type instance T2 a = Char class C3 a where -- equivalent to C1, C2 type T3 a type instance T3 a = Char data A = B deriving C1 deriving instance C2 A instance C3 A -- fails -- test1 :: T1 A -- test1 = 'x' -- fails -- test2 :: T2 A -- test2 = 'x' -- succeeds test3 :: T3 A test3 = 'x' }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10361> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 10