April 2017 - ghc-tickets

[GHC] #13621: Problems with injective type families
by GHC 27 Apr '17

27 Apr '17

#13621: Problems with injective type families -------------------------------------+------------------------------------- Reporter: Iceland_jack | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1 Keywords: | Operating System: Unknown/Multiple InjectiveFamilies | Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- {{{#!hs {-# LANGUAGE GADTs, TypeInType, DataKinds, TypeFamilyDependencies #-} import Data.Kind type family IB a = res | res -> a where IB Int = Bool IB Bool = Int type Cat k = k -> k -> Type data F :: Cat Type where F :: (a -> IB a) -> F a (IB a) data Exp :: Type -> Type where I :: Int -> Exp Int B :: Bool -> Exp Bool fmap' :: F a b -> (Exp a -> Exp b) fmap' (F f) = \case B b -> I (f b) I i -> B (f i) }}} I would have thought this should work as well, given that `IB` is injective: {{{#!hs data F :: Cat Type where F :: (IB a -> a) -> F (IB a) a }}} but it gives {{{#!hs fmap' :: F a b -> (Exp a -> Exp b) fmap' (F f) = \case B b -> I (f b) -- c.hs:33:13: error: -- • Could not deduce: b ~ Int -- from the context: a ~ IB b -- bound by a pattern with constructor: -- F :: forall a. (IB a -> a) -> F (IB a) a, -- in an equation for ‘fmap'’ -- at c.hs:32:8-10 -- or from: a ~ Bool -- bound by a pattern with constructor: B :: Bool -> Exp Bool, -- in a case alternative -- at c.hs:33:3-5 -- ‘b’ is a rigid type variable bound by -- the type signature for: -- fmap' :: forall a b. F a b -> Exp a -> Exp b -- at c.hs:31:10 -- • In the first argument of ‘I’, namely ‘(f b)’ -- In the expression: I (f b) -- In a case alternative: B b -> I (f b) -- • Relevant bindings include -- f :: IB b -> b (bound at c.hs:32:10) -- fmap' :: F a b -> Exp a -> Exp b (bound at c.hs:32:1) -- Failed, modules loaded: none. }}} But if we know that `a ~ Bool` (as we do from matching on `B`), and we know that `a ~ IB b` then by injectivity it should figure that `b ~ Int`? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13621> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

Re: [GHC] #3781: Improve inlining for local functions
by GHC 27 Apr '17

27 Apr '17

#3781: Improve inlining for local functions -------------------------------------+------------------------------------- Reporter: simonpj | Owner: (none) Type: bug | Status: new Priority: lowest | Milestone: Component: Compiler | Version: 6.12.1 Resolution: | Keywords: Inlining Operating System: Unknown/Multiple | Architecture: Type of failure: Runtime | Unknown/Multiple performance bug | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by simonpj): See also #3755, #7664 -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/3781#comment:16> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #13614: Rewrite rules not applied exhaustively when simplifying from plugin
by GHC 27 Apr '17

27 Apr '17

#13614: Rewrite rules not applied exhaustively when simplifying from plugin -------------------------------------+------------------------------------- Reporter: nomeata | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: GHC API | 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: -------------------------------------+------------------------------------- Consider this program: {{{ {-# OPTIONS_GHC -O -fplugin TestPlugin #-} module Test where foo :: Int -> Int foo = id {-# INLINE [0] foo #-} {-# RULES "foo1" [1] foo 1 = foo 2 "foo2" [1] foo 2 = foo 3 #-} fun :: Int -> Int -> Int fun = (+) {-# NOINLINE fun #-} test = foo 1 `fun` foo 2 }}} I would expect that one run of the simplifier in phase `1` will turn this into {{{ test = foo 3 `fun` foo 3 }}} I am using this plugin to test this: {{{ module TestPlugin where import System.Exit import Control.Monad import GhcPlugins import Simplify import CoreStats import SimplMonad import FamInstEnv import SimplEnv -- Plugin boiler plate plugin :: Plugin plugin = defaultPlugin { installCoreToDos = install } install :: [CommandLineOption] -> [CoreToDo] -> CoreM [CoreToDo] install _ (simpl:xs) = return $ simpl : pass : xs where pass = CoreDoPluginPass "Test" testPass -- The plugin testPass :: ModGuts -> CoreM ModGuts testPass guts = do let [expr] = [ e | NonRec v e <- mg_binds guts , occNameString (occName v) == "test" ] simplified_expression <- simplify guts expr putMsg $ text "Test" $$ nest 4 (hang (text "Before" <> colon) 4 (ppr expr)) $$ nest 4 (hang (text "After" <> colon) 4 (ppr simplified_expression)) liftIO $ exitFailure -- A simplifier simplify :: ModGuts -> CoreExpr -> CoreM CoreExpr simplify guts expr = do dflags <- getDynFlags let dflags' = dflags { ufVeryAggressive = True } us <- liftIO $ mkSplitUniqSupply 's' let sz = exprSize expr rule_base <- getRuleBase vis_orphs <- getVisibleOrphanMods let rule_base2 = extendRuleBaseList rule_base (mg_rules guts) let rule_env = RuleEnv rule_base2 vis_orphs (expr', _) <- liftIO $ initSmpl dflags' rule_env emptyFamInstEnvs us sz $ simplExpr (simplEnv 1) >=> simplExpr (simplEnv 1) $ expr return expr' simplEnv :: Int -> SimplEnv simplEnv p = mkSimplEnv $ SimplMode { sm_names = ["Test"] , sm_phase = Phase p , sm_rules = True , sm_inline = True , sm_eta_expand = True , sm_case_case = True } }}} But I get: {{{ $ ghc-head -package ghc Test.hs [1 of 2] Compiling TestPlugin ( TestPlugin.hs, TestPlugin.o ) [2 of 2] Compiling Test ( Test.hs, Test.o ) Test Before: fun (foo (GHC.Types.I# 1#)) (foo (GHC.Types.I# 2#)) After: fun (foo (GHC.Types.I# 2#)) (foo (GHC.Types.I# 3#)) }}} If I however compile this without the plugin, and look at what’s happening with `-dverbose-core2core`, I observe this: {{{ … test :: Int test = fun (foo (GHC.Types.I# 1#)) (foo (GHC.Types.I# 2#)) … ==================== Simplifier ==================== Max iterations = 4 SimplMode {Phase = 1 [main], inline, rules, eta-expand, case-of-case} … test :: Int test = fun (foo (GHC.Types.I# 3#)) (foo (GHC.Types.I# 3#)) … }}} So what am I doing wrong in my plugin? Any help is appreciated. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13614> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 10

[GHC] #13620: hsc2hs parses incorrectly '#ifdef' under '#{enum' in '--cross-compile' mode
by GHC 27 Apr '17

27 Apr '17

#13620: hsc2hs parses incorrectly '#ifdef' under '#{enum' in '--cross-compile' mode -------------------------------------+------------------------------------- Reporter: slyfox | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: hsc2hs | Version: 8.0.1 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: -------------------------------------+------------------------------------- The code is simplified version of something from Win32 package (https://github.com/haskell/win32/blob/master/System/Win32/SimpleMAPI.hsc#L56) {{{#!hs #{enum Int , , a = sizeof(int) #if 0 , b = sizeof(char) #endif } }}} {{{ $ hsc2hs b.hsc -o b.hs-native $ hsc2hs b.hsc -o b.hs-cross --cross-compile b.hsc:1 sizeof(int) #if 0 is not an integer }}} I'm not sure if it's a valid .hsc but having at least the same (valid or invalid) result in both modes would be nice. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13620> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #13619: hsc2hs parses incorrectly c99 style ("/// ...") comments in '--cross-compile' mode
by GHC 27 Apr '17

27 Apr '17

#13619: hsc2hs parses incorrectly c99 style ("/// ...") comments in '--cross- compile' mode -------------------------------------+------------------------------------- Reporter: slyfox | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: hsc2hs | Version: 8.0.1 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: -------------------------------------+------------------------------------- The code is simplified version of something from Win32 package (https://github.com/haskell/win32/blob/master/Graphics/Win32/GDI/Pen.hsc#L59) {{{#!hs -- a.hsc #{enum Int , // "hi" there! , a = sizeof(int) } }}} {{{ $ hsc2hs a.hsc -o a.hs-cross --cross-compile $ hsc2hs a.hsc -o a.hs-native }}} {{{#!diff --- a.hs-cross 2017-04-26 21:56:56.157323964 +0100 +++ a.hs-native 2017-04-26 21:57:00.753311717 +0100 @@ -2,3 +2,3 @@ a :: Int -a = // "hi" there! 4 +a = 4 }}} I'm not sure if it's a valid .hsc but having at least the same (valid or invalid) result in both modes would be nice. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13619> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #13616: Old file used when later calls to GHC omit -dynamic-too
by GHC 27 Apr '17

27 Apr '17

#13616: Old file used when later calls to GHC omit -dynamic-too -------------------------------------+------------------------------------- Reporter: nomeata | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1 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: -------------------------------------+------------------------------------- This may possibly only hurt few people who develop Plugins, but it is still quite annoying: If `ghc`, in `--make` mode, recompiles a plugin that it is going to use, but without `-dynamic-too`, it will create new `.dyn` and `.o` files, but it will happily use the old `.dyn_hs` or `.dyn_o` files that are still lying around. I guess the desired behavior is: If it determines that it needs the dynamically compiled version of some module, and it is recompiling the module, then it should complain instead of later either ungracefully falling over the missing `.dyn_o` file or – worse – using an old one. This script reproduces the issue: {{{ #!/bin/bash rm -f *.o *.hi *.dyn_o *.dyn_hi cat > Plugin.hs <<__END__ module Plugin where import GhcPlugins plugin :: Plugin plugin = defaultPlugin { installCoreToDos = install } install :: [CommandLineOption] -> [CoreToDo] -> CoreM [CoreToDo] install _ _ = error "First version" __END__ cat > Code.hs <<__END__ {-# OPTIONS_GHC -O -fplugin Plugin #-} module Code where __END__ echo "No dynamic file found (fine, but the error message could be more helpful)" ghc -package ghc Code.hs -fforce-recomp echo "Compiling with -dynamic-too" ghc -dynamic-too -package ghc Code.hs -fforce-recomp cat > Plugin.hs <<__END__ module Plugin where import GhcPlugins plugin :: Plugin plugin = defaultPlugin { installCoreToDos = install } install :: [CommandLineOption] -> [CoreToDo] -> CoreM [CoreToDo] install _ _ = error "Second version" __END__ echo "Recompiling without -dynamic-too, it still uses the old file" ghc -package ghc Code.hs -fforce-recomp echo "Recompiling with -dynamic-too, now it works" ghc -dynamic-too -package ghc Code.hs -fforce-recomp }}} And here is the slightly redacted output: {{{ $ ./repro.sh No dynamic file found (fine, but the error message could be more helpful) [1 of 2] Compiling Plugin ( Plugin.hs, Plugin.o ) [2 of 2] Compiling Code ( Code.hs, Code.o ) <no location info>: fatal: cannot find object file ‘./Plugin.dyn_o’ while linking an interpreted expression Compiling with -dynamic-too [1 of 2] Compiling Plugin ( Plugin.hs, Plugin.o ) [2 of 2] Compiling Code ( Code.hs, Code.o ) ghc: panic! (the 'impossible' happened) First version Recompiling without -dynamic-too, it still uses the old file [1 of 2] Compiling Plugin ( Plugin.hs, Plugin.o ) [2 of 2] Compiling Code ( Code.hs, Code.o ) ghc: panic! (the 'impossible' happened) First version Recompiling with -dynamic-too, now it works [1 of 2] Compiling Plugin ( Plugin.hs, Plugin.o ) [2 of 2] Compiling Code ( Code.hs, Code.o ) ghc: panic! (the 'impossible' happened) Second version }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13616> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #13598: role annotation for newtype (partially?) ignored?
by GHC 26 Apr '17

26 Apr '17

#13598: role annotation for newtype (partially?) ignored? -------------------------------------+------------------------------------- Reporter: edsko | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 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: -------------------------------------+------------------------------------- A constraint `Coercible o (T a)` is ambiguous if the role of `a` is representational (provided `a` is not otherwise constrained, of course), but unambiguous if `a` is a phantom type: {{{#!hs {-# LANGUAGE FlexibleContexts #-} {-# LANGUAGE RoleAnnotations #-} module CoerceTest where import Data.Coerce type role A phantom data A a = MkA Int type role B representational data B a = MkB Int -- accepted by the type checker (as it should be) fooA :: Coercible o (A a) => o -> () fooA _ = () -- rejected by the type checker (as it should be) with -- "Couldn't match representation of type ‘a0’ with that of ‘a’" -- fooB :: Coercible o (B a) => o -> () -- fooB _ = () }}} However, for `newtype`s something odd happens: {{{#!hs type role C representational newtype C a = MkC Int -- accepted by the type checker (but should not be) fooC :: Coercible o (C a) => o -> () fooC _ = () }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13598> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #13583: configure.ac: handle AR=, NM= and other flags as arguments to ./configure
by GHC 26 Apr '17

26 Apr '17

#13583: configure.ac: handle AR=, NM= and other flags as arguments to ./configure -------------------------------------+------------------------------------- Reporter: slyfox | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1 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: -------------------------------------+------------------------------------- A followup on comments around https://phabricator.haskell.org/rGHC79848f18805a The following does not work currently: {{{ $ ./configure \ --enable-unregisterised \ --enable-bfd-debug \ NM=gcc-nm \ AR=gcc-ar \ RANLIB=gcc-ranlib ... ar : /usr/bin/ar ld : /usr/bin/ld nm : /usr/bin/nm objdump : /usr/bin/objdump ranlib : /usr/bin/ranlib }}} The workaround is to use {{{ $ ./configure --enable-unregisterised \ --with-nm=gcc-nm \ --with-ar=gcc-ar \ --with-ranlib=gcc-ranlib }}} The plan is to make it work (perhaps by using '''AC_PROG_AR'''?). -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13583> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #13097: Num a => Num (Down a)
by GHC 26 Apr '17

26 Apr '17

#13097: Num a => Num (Down a) -------------------------------------+------------------------------------- Reporter: Iceland_jack | Owner: Type: feature | Status: new request | Priority: normal | Milestone: Component: | Version: 8.0.1 libraries/base | 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: -------------------------------------+------------------------------------- Let's us write {{{#!hs >>> deriving instance Num a => Num (Down a) >>> (<) 10 20 True >>> (<) @(Down _) 10 20 False }}} With #12363 {{{#!hs >>> 10 < @(Down _) 20 False }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13097> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 8

[GHC] #13420: Bizarre pretty-printing of closed type families in GHCi
by GHC 26 Apr '17

26 Apr '17

#13420: Bizarre pretty-printing of closed type families in GHCi -------------------------------------+------------------------------------- Reporter: RyanGlScott | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: GHCi | Version: 8.0.2 Keywords: TypeFamilies | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Load this code: {{{#!hs {-# LANGUAGE TypeFamilies #-} module Bug where type family F a where F [Int] = Bool F [a] = Double F (a b) = Char }}} into GHCi 8.0.1, 8.0.2, or HEAD, and run `:i F`. You'll see this: {{{ $ /opt/ghc/8.0.2/bin/ghci Bug.hs GHCi, version 8.0.2: http://www.haskell.org/ghc/ :? for help Loaded GHCi configuration from /home/rgscott/.ghci [1 of 1] Compiling Bug ( Bug.hs, interpreted ) Ok, modules loaded: Bug. λ> :i F type family F a :: * where F [Int] = Bool [a] F [a] = Double [b, (a :: * -> *)] F (a b) = Char -- Defined at Bug.hs:4:1 }}} Compare this with GHCi 7.10.3, where the pretty-printing is far more sane: {{{ $ /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 ) Ok, modules loaded: Bug. λ> :i F type family F a :: * where F [Int] = Bool F [a] = Double F (a b) = Char -- Defined at Bug.hs:4:1 }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13420> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5