August 2018 - ghc-tickets

[GHC] #15480: Rename SigTv
by GHC 17 Aug '18

17 Aug '18

#15480: Rename SigTv -------------------------------------+------------------------------------- Reporter: nomeata | Owner: (none) Type: task | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.5 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: -------------------------------------+------------------------------------- After #15050 and [changeset:"4d91cabcd5e3c603997d9876f6d30204a9b029c6/ghc" 4d91cabc/ghc] it seems odd to call `SigTv` for the kind of type variables that used to be used in type signatures in patterns, but are now used in other places (kind signatures, partial type signatures). Also, `SigTv` may be associated with sigma. At some point I assumed that `SigTv` are type variables that may unify with polytypes, while `TauTv` are those that unify with monotypes. As a service to our future self, I’d like to rename `SigTv`. Suggestions? Best one I have is `TvTv` or `TyVarTv` (for a type variable that stands for a type variable). -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15480> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #8944: Warn instead of stopping on misplaced Haddock comments
by GHC 17 Aug '18

17 Aug '18

#8944: Warn instead of stopping on misplaced Haddock comments -------------------------------------+------------------------------------- Reporter: Fuuzetsu | Owner: Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler (Parser) | Version: 7.9 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: None/Unknown Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | -------------------------------------+------------------------------------- Given a simple module like {{{ module H where data F = F () -- ^ Comment for first type argument () }}} and trying to run Haddock on it, we'll get back {{{ /tmp/H.hs:4:12: parse error on input ‘(’ }}} The error message is rather sub-par in this scenario. It'd be great if we could instead print a warning saying that a Haddock comment is unexpected in that position and then continue on parsing. Best case scenario would be a more informative message such as “Documenting each constructor argument is not supported” but that might be quite a bit harder. Filing on GHC Trac as it's the parser that needs changing I believe. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8944> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 11

[GHC] #15521: Provide a strict version of sum
by GHC 16 Aug '18

16 Aug '18

#15521: Provide a strict version of sum -------------------------------------+------------------------------------- Reporter: dnadales | Owner: (none) Type: feature | Status: new request | Priority: normal | Milestone: 8.6.1 Component: Prelude | Version: 8.4.3 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: -------------------------------------+------------------------------------- When adding huge list of numbers, a strict version of `sum` is preferred to avoid an unnecessary use of memory. A strict version of `sum` can be easily written using `foldl'` but it'd be nicer if a function such as: {{{#!hs sum' = foldl' (+) 0 }}} would be provided in the prelude already. Does this makes sense? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15521> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 8

[GHC] #15453: Bug in opt_trans_rule in OptCoercion
by GHC 16 Aug '18

16 Aug '18

#15453: Bug in opt_trans_rule in OptCoercion -------------------------------------+------------------------------------- Reporter: ningning | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.6.1 Component: Compiler | Version: 8.4.3 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: -------------------------------------+------------------------------------- The ForAllCo case in the definition of opt_trans_rule in OptCoercion is not right: {{{#!hs push_trans tv1 eta1 r1 tv2 eta2 r2 = fireTransRule "EtaAllTy" co1 co2 $ mkForAllCo tv1 (opt_trans is eta1 eta2) (opt_trans is' r1 r2') where is' = is `extendInScopeSet` tv1 r2' = substCoWithUnchecked [tv2] [TyVarTy tv1] r2 -- ill-kinded! }}} Given {{{co1;co2}}}, where {{{#!hs co1 = \/ tv1 : eta1. r1 co2 = \/ tv2 : eta2. r2 }}} We would like optimize the transitivity coercion. I think what we want is {{{#!hs \/tv1 : (eta1;eta2). (r1; r2[tv2 |-> tv1 |> eta1]) }}} Namely it should be {{{#!hs r2' = substCoWithUnchecked [tv2] [mkCastTy (TyVarTy tv1) eta1] r2 }}} If there is any program that could hit the bug it would be better, as we can add it to test cases. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15453> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 18

[GHC] #15472: GHC HEAD type inference regression post-"Remove decideKindGeneralisationPlan"
by GHC 16 Aug '18

16 Aug '18

#15472: GHC HEAD type inference regression post-"Remove decideKindGeneralisationPlan" -------------------------------------+------------------------------------- Reporter: RyanGlScott | Owner: (none) Type: bug | Status: new Priority: highest | Milestone: 8.8.1 Component: Compiler | Version: 8.5 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: -------------------------------------+------------------------------------- Commit c955a514f033a12f6d0ab0fbacec3e18a5757ab5 (`Remove decideKindGeneralisationPlan`) causes the `singletons` library to no longer compile. Here is as minimal of an example as I can muster: {{{#!hs {-# LANGUAGE GADTs #-} {-# LANGUAGE RankNTypes #-} {-# LANGUAGE ScopedTypeVariables #-} {-# LANGUAGE TypeApplications #-} {-# LANGUAGE TypeFamilies #-} {-# LANGUAGE TypeInType #-} {-# LANGUAGE TypeOperators #-} {-# LANGUAGE UndecidableInstances #-} module Bug (sPermutations) where import Data.Kind data TyFun :: Type -> Type -> Type type a ~> b = TyFun a b -> Type infixr 0 ~> type family Apply (f :: k1 ~> k2) (x :: k1) :: k2 data family Sing :: forall k. k -> Type data instance Sing :: forall a. [a] -> Type where SNil :: Sing '[] SCons :: Sing x -> Sing xs -> Sing (x:xs) newtype instance Sing (f :: k1 ~> k2) = SLambda { applySing :: forall t. Sing t -> Sing (Apply f t) } type SingFunction1 f = forall t. Sing t -> Sing (Apply f t) singFun1 :: forall f. SingFunction1 f -> Sing f singFun1 f = SLambda f type SingFunction2 f = forall t. Sing t -> SingFunction1 (Apply f t) singFun2 :: forall f. SingFunction2 f -> Sing f singFun2 f = SLambda (\x -> singFun1 (f x)) type family Undefined :: k where {} sUndefined :: a sUndefined = undefined type family LetGo k z x ys where LetGo k z x '[] = z LetGo k z x (y ': ys) = Apply (Apply k y) (LetGo k z x ys) type family Foldr (k :: a ~> b ~> b) (x :: b) (xs :: [a]) :: b where Foldr k z x = LetGo k z x x sFoldr :: forall (t1 :: a ~> b ~> b) (t2 :: b) (t3 :: [a]). Sing t1 -> Sing t2 -> Sing t3 -> Sing (Foldr t1 t2 t3 :: b) sFoldr (sK :: Sing k) (sZ :: Sing z) (sX :: Sing x) = let sGo :: forall arg. Sing arg -> Sing (LetGo k z x arg) sGo SNil = sZ sGo (SCons (sY :: Sing y) (sYs :: Sing ys)) = sK `applySing` sY `applySing` sGo sYs in sGo sX type family LetInterleave xs0 t ts is (a1 :: [a]) (a2 :: [[a]]) :: [[a]] where LetInterleave xs0 t ts is a1 a2 = Undefined a1 a2 data LetInterleaveSym5 l_ahko l_ahkp l_ahkq l_ahkr (l_ahks :: [a]) :: [[a]] ~> [[a]] type instance Apply (LetInterleaveSym5 l_ahko l_ahkp l_ahkq l_ahkr l_ahks) l_ahkn = LetInterleave l_ahko l_ahkp l_ahkq l_ahkr l_ahks l_ahkn data LetInterleaveSym4 l_ahkv l_ahkw l_ahkx l_ahky :: forall a. [a] ~> [[a]] ~> [[a]] type instance Apply (LetInterleaveSym4 l_ahkv l_ahkw l_ahkx l_ahky) l_ahku = LetInterleaveSym5 l_ahkv l_ahkw l_ahkx l_ahky l_ahku data LetInterleaveSym3 l_ahkB l_ahkC l_ahkD l_ahkA type instance Apply (LetInterleaveSym3 l_ahkB l_ahkC l_ahkD) l_ahkA = LetInterleaveSym4 l_ahkB l_ahkC l_ahkD l_ahkA data LetInterleaveSym2 l_ahkG l_ahkH l_ahkF type instance Apply (LetInterleaveSym2 l_ahkG l_ahkH) l_ahkF = LetInterleaveSym3 l_ahkG l_ahkH l_ahkF data LetInterleaveSym1 l_ahkK l_ahkJ type instance Apply (LetInterleaveSym1 l_ahkK) l_ahkJ = LetInterleaveSym2 l_ahkK l_ahkJ data LetInterleaveSym0 l_ahkM type instance Apply LetInterleaveSym0 l_ahkM = LetInterleaveSym1 l_ahkM type family LetPerms xs0 a1 a2 where LetPerms xs0 '[] _ = '[] LetPerms xs0 (t ': ts) is = Foldr (LetInterleaveSym4 xs0 t ts is) (LetPerms xs0 ts (t ': is)) (Permutations is) {- $(singletons [d| permutations :: forall a. [a] -> [[a]] permutations xs0 = xs0 : perms xs0 [] where perms [] _ = [] perms (t:ts) is = foldr interleave (perms ts (t:is)) (permutations is) where interleave :: [a] -> [[a]] -> [[a]] interleave = undefined |]) -} type family Permutations (xs0 :: [a]) :: [[a]] where Permutations xs0 = xs0 ': LetPerms xs0 xs0 '[] sPermutations :: forall (t1 :: [a]). Sing t1 -> Sing (Permutations t1 :: [[a]]) sPermutations (sXs0 :: Sing xs0) = let sPerms :: forall arg1 arg2. Sing arg1 -> Sing arg2 -> Sing (LetPerms xs0 arg1 arg2) sPerms SNil _ = SNil sPerms (SCons (sT :: Sing t) (sTs :: Sing ts)) (sIs :: Sing is) = let sInterleave :: forall (t2 :: [a]) (t3 :: [[a]]). Sing t2 -> Sing t3 -> Sing (LetInterleave xs0 t ts is t2 t3 :: [[a]]) sInterleave (sA1 :: Sing a1) (sA2 :: Sing a2) = sUndefined sA1 sA2 in sFoldr (singFun2 @(LetInterleaveSym4 xs0 t ts is) sInterleave) (sPerms sTs (sT `SCons` sIs)) (sPermutations sIs) in sXs0 `SCons` sPerms sXs0 SNil }}} After that commit, this program (which previously typechecked) now errors with: {{{ $ ~/Software/ghc4/inplace/bin/ghc-stage2 Bug.hs [1 of 1] Compiling Bug ( Bug.hs, Bug.o ) Bug.hs:105:66: error: • Could not deduce: t2 ~~ t30 from the context: arg1 ~ (x : xs) bound by a pattern with constructor: SCons :: forall a (x :: a) (xs :: [a]). Sing x -> Sing xs -> Sing (x : xs), in an equation for ‘sPerms’ at Bug.hs:99:17-53 ‘t2’ is a rigid type variable bound by a type expected by the context: SingFunction2 (LetInterleaveSym4 t1 x xs arg2) at Bug.hs:105:24-76 Expected type: Sing t -> Sing t2 -> Sing (Apply (Apply (LetInterleaveSym4 t1 x xs arg2) t) t2) Actual type: Sing t20 -> Sing t30 -> Sing (LetInterleave t1 x xs arg2 t20 t30) • In the second argument of ‘singFun2’, namely ‘sInterleave’ In the first argument of ‘sFoldr’, namely ‘(singFun2 @(LetInterleaveSym4 xs0 t ts is) sInterleave)’ In the expression: sFoldr (singFun2 @(LetInterleaveSym4 xs0 t ts is) sInterleave) (sPerms sTs (sT `SCons` sIs)) (sPermutations sIs) • Relevant bindings include sInterleave :: forall (t2 :: [a1]) (t3 :: [[a1]]). Sing t2 -> Sing t3 -> Sing (LetInterleave t1 x xs arg2 t2 t3) (bound at Bug.hs:103:17) sIs :: Sing arg2 (bound at Bug.hs:99:57) sTs :: Sing xs (bound at Bug.hs:99:39) sT :: Sing x (bound at Bug.hs:99:24) sPerms :: Sing arg1 -> Sing arg2 -> Sing (LetPerms t1 arg1 arg2) (bound at Bug.hs:98:9) sXs0 :: Sing t1 (bound at Bug.hs:94:16) (Some bindings suppressed; use -fmax-relevant-binds=N or -fno-max- relevant-binds) | 105 | in sFoldr (singFun2 @(LetInterleaveSym4 xs0 t ts is) sInterleave) | ^^^^^^^^^^^ }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15472> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 11

[GHC] #15520: Invalid warning Defined but not used
by GHC 16 Aug '18

16 Aug '18

#15520: Invalid warning Defined but not used -------------------------------------+------------------------------------- Reporter: rik | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.6.1 Component: Compiler | Version: 8.4.3 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: -------------------------------------+------------------------------------- This code gives warning: Defined but not used: data constructor ‘:+++’ {{{#!hs data U = Text :+++ Text x :: U x = "ss" :+++ "ss" }}} Also In prefix notation . -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15520> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #15512: Rewrite rules should be able to produce custom compiler errors
by GHC 16 Aug '18

16 Aug '18

#15512: Rewrite rules should be able to produce custom compiler errors -------------------------------------+------------------------------------- Reporter: ChaiTRex | Owner: (none) Type: feature | Status: new request | Priority: normal | Milestone: 8.6.1 Component: Compiler | Version: 8.4.3 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: -------------------------------------+------------------------------------- It would be nice if rewrite rules could produce compiler errors with custom messages when a certain combination of functions can only produce bottom or some other suitably undesirable outcome. For example, `length . cycle` is either never going to terminate or is going to give an error message (''e''.''g''., if the input list is empty). It would be nice to be able to tell the programmer that at compile time. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15512> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #15528: Match is missing a Data instance
by GHC 16 Aug '18

16 Aug '18

#15528: Match is missing a Data instance -------------------------------------+------------------------------------- Reporter: mpickering | Owner: (none) Type: task | Status: new Priority: normal | Milestone: 8.6.1 Component: Compiler | Version: 8.4.3 Keywords: newcomer | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- If I try to use `showAstData` on a `Match` then GHC complains that there is no Data instance. {{{ No instance for (Data (Expr.Match p (Expr.LHsExpr p))) }}} It seems that all the constituent parts of `Match` have a `Data` instance so `Match` should as well. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15528> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1

[GHC] #15149: Identical distinct type family fields miscompiled
by GHC 16 Aug '18

16 Aug '18

#15149: Identical distinct type family fields miscompiled -------------------------------------+------------------------------------- Reporter: NeilMitchell | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.6.1 Component: Compiler | Version: 8.4.2 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: GHC rejects Unknown/Multiple | valid program Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Given the code: {{{#!hs -- An.hs {-# LANGUAGE TypeFamilies #-} module An where data family An c :: * -- AnInt.hs {-# LANGUAGE TypeFamilies #-} module AnInt where import An data instance An Int = AnInt {an :: Int} deriving Show -- AnDouble.hs {-# LANGUAGE TypeFamilies #-} module AnDouble where import An data instance An Double = AnDouble {an :: Double} deriving Show -- Main.hs {-# LANGUAGE DisambiguateRecordFields #-} module Main where import AnInt import AnDouble main = print (AnDouble{an=1}, AnInt{an=1}) }}} I would expect this code to work. In reality it fails at runtime with GHC 8.2.2: {{{ Main.hs:4:15-28: warning: [-Wmissing-fields] * Fields of `AnDouble' not initialised: an * In the expression: AnDouble {an = 1} In the first argument of `print', namely `(AnDouble {an = 1}, AnInt {an = 1})' In the expression: print (AnDouble {an = 1}, AnInt {an = 1}) | 6 | main = print (AnDouble{an=1}, AnInt{an=1}) | ^^^^^^^^^^^^^^ *** Exception: Main.hs:4:15-28: Missing field in record construction an }}} And fails at compile time in GHC 8.4.2: {{{ Main.hs:4:31-41: error: * Constructor `AnInt' does not have field `an' * In the expression: AnInt {an = 1} In the first argument of `print', namely `(AnDouble {an = 1}, AnInt {an = 1})' In the expression: print (AnDouble {an = 1}, AnInt {an = 1}) | 6 | main = print (AnDouble{an=1}, AnInt{an=1}) | ^^^^^^^^^^^ }}} This code was extracted from a real example, where this bug is pretty fatal, as I haven't been able to find any workarounds (without just avoiding clashing record fields). -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15149> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 14

Re: [GHC] #7535: Using -with-rtsopts=-N should fail unless -threaded is also specified
by GHC 16 Aug '18

16 Aug '18

#7535: Using -with-rtsopts=-N should fail unless -threaded is also specified -------------------------------------+------------------------------------- Reporter: mhoermann | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.6.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: 4243 | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Changes (by osa1): * cc: osa1 (added) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7535#comment:7> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0