July 2017 - ghc-tickets

[GHC] #9661: Branchless ==# is compiled to branchy code
by GHC 08 Sep '17

08 Sep '17

#9661: Branchless ==# is compiled to branchy code -------------------------------------+------------------------------------- Reporter: dfeuer | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.9 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Runtime Blocked By: | performance bug Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- This started as a comment on #6135, but Jan Stolarek requested that I file a separate report. The branchless tests `<#`, `>#`, `<=#`, and `>=#`, and their wordy cousins, seem to work properly, but `==#` and `eqWord#` don't. If I write {{{#!hs foo x = tagToEnum# ((x <# 3#) `orI#` (x ># 100#) `orI#` (x ==# 12#) `orI#` (x ==# 15#)) }}} I get (in 7.8.3 and in 7.9) {{{#!hs IsDigit.foo :: GHC.Prim.Int# -> GHC.Types.Bool [GblId, Arity=1, Caf=NoCafRefs, Str=DmdType <S,1*U>, Unf=Unf{Src=InlineStable, TopLvl=True, Value=True, ConLike=True, WorkFree=True, Expandable=True, Guidance=ALWAYS_IF(arity=1,unsat_ok=True,boring_ok=False) Tmpl= \ (x3_a2if [Occ=Once!] :: GHC.Prim.Int#) -> case x3_a2if of wild_Xe { __DEFAULT -> GHC.Prim.tagToEnum# @ GHC.Types.Bool (GHC.Prim.orI# (GHC.Prim.<# wild_Xe 3) (GHC.Prim.># wild_Xe 100)); 12 -> GHC.Types.True; 15 -> GHC.Types.True }}] IsDigit.foo = \ (x3_a2if :: GHC.Prim.Int#) -> case x3_a2if of wild_Xe { __DEFAULT -> GHC.Prim.tagToEnum# @ GHC.Types.Bool (GHC.Prim.orI# (GHC.Prim.<# wild_Xe 3) (GHC.Prim.># wild_Xe 100)); 12 -> GHC.Types.True; 15 -> GHC.Types.True }}} and branching assembly to match. Jan Stolarek indicates this is probably a problem with constant folding. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9661> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 30

[GHC] #10365: Make Semigroup as a superclass of Monoid
by GHC 07 Sep '17

07 Sep '17

#10365: Make Semigroup as a superclass of Monoid -------------------------------------+------------------------------------- Reporter: gidyn | Owner: Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.10.1 Component: | Operating System: Unknown/Multiple libraries/base | Type of failure: None/Unknown Keywords: | Blocked By: Architecture: | Related Tickets: Unknown/Multiple | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- See http://haskell.1045720.n5.nabble.com/Proposal-Make-Semigroup-as-a -superclass-of-Monoid-td5767835.html -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10365> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 45

[GHC] #9797: Investigate rewriting `>>=` to `*>` or `>>` for appropriate types
by GHC 05 Sep '17

05 Sep '17

#9797: Investigate rewriting `>>=` to `*>` or `>>` for appropriate types -------------------------------------+------------------------------------- Reporter: dfeuer | Owner: dfeuer Type: task | Status: new Priority: normal | Milestone: 7.10.1 Component: Core Libraries | Version: 7.9 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Runtime Blocked By: | performance bug Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- When we see `m >>= (\_ -> n)` with a type that admits an optimized `>>` or `<$`, we should try to take advantage of that. I don't currently know if this applies to any types under "GHC HQ" control. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9797> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #10328: Control.Monad exports lead to weird Haddocks
by GHC 05 Sep '17

05 Sep '17

#10328: Control.Monad exports lead to weird Haddocks -------------------------------------+------------------------------------- Reporter: dfeuer | Owner: ekmett Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.10.1 Component: Core | Operating System: Unknown/Multiple Libraries | Type of failure: Documentation Keywords: | bug Architecture: | Blocked By: Unknown/Multiple | Related Tickets: Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- `Control.Monad` exports, and therefore documents, the `Functor` class (but only the `fmap` member) and the `Monad` class. It does not export, and therefore does not document, the `Applicative` class. This is all pretty weird. Since `Functor` is in the prelude (and probably most/all alternate preludes), is there actually a reason to export it from `Control.Monad`? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10328> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #9756: Warn about unnecessary unsafeCoerce
by GHC 05 Sep '17

05 Sep '17

#9756: Warn about unnecessary unsafeCoerce -------------------------------------+------------------------------------- Reporter: dfeuer | Owner: Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler (Type | Version: 7.9 checker) | Operating System: Keywords: | Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: Difficulty: Unknown | None/Unknown Blocked By: | Test Case: Related Tickets: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- `unsafeCoerce` has been around a lot longer than `coerce`, so older code tends to use it even if `coerce` could do the job. It would be nice if the type checker could produce a warning when it detected such a situation. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9756> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #13779: Add more signature suppression control for dumps
by GHC 05 Sep '17

05 Sep '17

#13779: Add more signature suppression control for dumps -------------------------------------+------------------------------------- Reporter: dfeuer | Owner: (none) Type: feature | Status: new request | Priority: normal | Milestone: 8.4.1 Component: Compiler | Version: 8.2.1-rc2 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: -------------------------------------+------------------------------------- I tend to find some type signatures essential when reading Core. However, `Generic`-heavy code can produce extremely large signatures. I suspect one or both of the following additional options could help: 1. An option to suppress non-top-level signatures, leaving top-level ones alone. 2. An option to suppress phantom type arguments in signatures. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13779> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #13182: Rethinking dataToTag#
by GHC 05 Sep '17

05 Sep '17

#13182: Rethinking dataToTag# -------------------------------------+------------------------------------- Reporter: dfeuer | Owner: Type: task | Status: new Priority: normal | Milestone: 8.2.1 Component: Compiler | Version: 8.1 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: -------------------------------------+------------------------------------- We currently use a primop `dataToTag#` that requires some careful handling because it can only be applied to something that has already been evaluated. We (now) mark it `can_fail` to prevent it from floating out, and then add a fix-up pass in CorePrep just in case things went wrong. Could we do something simpler? Suppose we had {{{#!hs -- The real primop, preferably not exported from *anywhere*, but at least documented as never to be used outside GHC. secretDataToTag# :: a -> Int# -- A wired-in safe version dataToTag# :: a -> Int# dataToTag# !a = secretDataToTag# a {-# NOINLINE dataToTag# #-} }}} Then `dataToTag#` could be inlined in CorePrep, just like `runRW#`. === Questions === How would this interact with float out? It should be perfectly safe to float out when its argument is known to be forced, but it seems likely we'll run into let/app invariant failures. What RULES do we (or should we) have for `dataToTag#`? How will this change affect them? My vague recollection is that we don't have any, but it would be pretty nice to apply something equivalent to case-of-known- constructor. Whenever `dataToTag#` is applied to a known constructor, we should be able to evaluate it. The earlier we do this the better it will inform simplification. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13182> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #13985: GHC 8.0 regression: ‘k’ is not in scope during type checking, but it passed the renamer
by GHC 05 Sep '17

05 Sep '17

#13985: GHC 8.0 regression: ‘k’ is not in scope during type checking, but it passed the renamer -------------------------------------+------------------------------------- Reporter: RyanGlScott | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1 (Type checker) | Keywords: TypeFamilies | Operating System: Unknown/Multiple Architecture: | Type of failure: Compile-time Unknown/Multiple | crash or panic Test Case: | Blocked By: Blocking: | Related Tickets: #13738 Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- {{{#!hs {-# LANGUAGE PolyKinds #-} {-# LANGUAGE RankNTypes #-} {-# LANGUAGE TypeFamilies #-} module Bug where import Data.Proxy data family Fam data instance Fam = MkFam (forall (a :: k). Proxy a) }}} On GHC 8.0.1, 8.0.2, 8.2.1, and HEAD, this fails with a GHC internal error: {{{ $ /opt/ghc/8.2.1/bin/ghci Bug.hs GHCi, version 8.2.0.20170704: http://www.haskell.org/ghc/ :? for help Loaded GHCi configuration from /home/rgscott/.ghci [1 of 1] Compiling Bug ( Bug.hs, interpreted ) Bug.hs:9:41: error: • GHC internal error: ‘k’ is not in scope during type checking, but it passed the renamer tcl_env of environment: [r1vy :-> APromotionErr FamDataConPE] • In the kind ‘k’ In the type ‘(forall (a :: k). Proxy a)’ In the definition of data constructor ‘MkFam’ | 9 | data instance Fam = MkFam (forall (a :: k). Proxy a) | ^ }}} This is a regression, since on GHC 7.10.3, it did not crash: {{{ $ /opt/ghc/7.10.3/bin/ghci Bug.hs GHCi, version 7.10.3: http://www.haskell.org/ghc/ :? for help [1 of 1] Compiling Bug ( Bug.hs, interpreted ) Bug.hs:9:21: Data constructor ‘MkFam’ has existential type variables, a context, or a specialised result type MkFam :: forall (k :: BOX). (forall (k :: BOX) (a :: k). Proxy a) -> Fam (Use ExistentialQuantification or GADTs to allow this) In the definition of data constructor ‘MkFam’ In the data instance declaration for ‘Fam’ }}} This smells like #13738, but no `TypeApplications` are involved here, so I decided to open a separate ticket to be safe. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13985> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #9574: GHC Panic: No Skolem Info
by GHC 05 Sep '17

05 Sep '17

#9574: GHC Panic: No Skolem Info -------------------------------------+------------------------------------- Reporter: ian_mi | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.9 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Blocked By: | None/Unknown Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- GHC 7.8.3 and from git both panic with Couldn't match kind ‘o1’ with ‘(,) o o’ ‘o1’ is untouchable inside the constraints (Object (c :><: c) a) bound by a type expected by the context: Object (c :><: c) a => Tagged a (c (FMap (ProductF c) a) (FMap (Proj2 c c) a)) at Product.hs:35:9-15ghc: panic! (the 'impossible' happened) (GHC version 7.9.20140908 for x86_64-unknown-linux): No skolem info: o1_a6XS I have not yet attempted to reduce the problem at all. The offending code is the use of fmap here: {{{#!hs proj2 :: forall (c :: o -> o -> *). ProductCategory c => NatTr (c :><: c) c (ProductF c) (Proj2 c c) proj2 = NatTr t where NatTr t = fmap (CompF Proj2 IdentityF) e e :: NatTr (c :><: c) (c :><: c) (Comp ('KProxy :: KProxy o) (Diag c) (ProductF c)) (IdentityF (c :><: c)) e = counit }}} fmap is defined as {{{#!hs fmap :: forall f (a :: o1) (b :: o1). Functor f ('KProxy :: KProxy (o1 -> o2)) => f -> Domain f a b -> Codomain f (FMap f a :: o2) (FMap f b :: o2) fmap _ = proxy morphMap (Proxy :: Proxy f) }}} while CompF is defined by {{{#!hs data CompF f g (k :: KProxy (o3 -> o4, o1 -> o2)) where CompF :: (Functor f ('KProxy :: KProxy (o3 -> o4)), Functor g ('KProxy :: KProxy (o1 -> o2))) => f -> g -> CompF f g ('KProxy :: KProxy (o3 -> o4, o1 -> o2)) }}} This makes some sense because there is a functional dependency between the functor and the kind proxy. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9574> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 9

[GHC] #13988: GADT constructor with kind equality constraint quantifies unused existential type variables
by GHC 05 Sep '17

05 Sep '17

#13988: GADT constructor with kind equality constraint quantifies unused existential type variables -------------------------------------+------------------------------------- Reporter: RyanGlScott | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1 (Type checker) | Keywords: TypeInType | Operating System: Unknown/Multiple Architecture: | Type of failure: Incorrect result Unknown/Multiple | at runtime Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Reproducible with GHC 8.0.1., 8.0.2, 8.2.1, or HEAD (but I'll use GHC 8.2.1 to show the results of `:type +v`): {{{ $ /opt/ghc/8.2.1/bin/ghci GHCi, version 8.2.0.20170704: http://www.haskell.org/ghc/ :? for help Loaded GHCi configuration from /home/rgscott/.ghci λ> :set -XTypeInType -XGADTs -fprint-explicit-foralls λ> import Data.Kind λ> data Foo (a :: k) where MkFoo :: (k ~ Type) => Foo (a :: k) λ> :type +v MkFoo MkFoo :: forall k1 (a :: k1) k2 (k3 :: k2). k1 ~ * => Foo a }}} Note that `MkFoo` quantifies two extra existential type variables, `k2` and `k3` which are completely unused! Note that this does not occur if the `MkFoo` constructor uses an explicit `forall`: {{{ λ> :set -XScopedTypeVariables λ> data Foo (a :: k) where MkFoo :: forall k (a :: k). (k ~ Type) => Foo (a :: k) λ> :type +v MkFoo MkFoo :: forall k (a :: k). k ~ * => Foo a }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13988> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 8