October 2017 - ghc-tickets

[GHC] #11549: Add -fshow-runtime-rep
by GHC 29 Jul '18

29 Jul '18

#11549: Add -fshow-runtime-rep -------------------------------------+------------------------------------- Reporter: goldfire | Owner: goldfire Type: bug | Status: new Priority: highest | Milestone: 8.0.1 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: -------------------------------------+------------------------------------- As discussed in several interwoven threads ([https://mail.haskell.org/pipermail/ghc-devs/2016-February/011268.html original], [https://mail.haskell.org/pipermail/haskell- cafe/2016-February/122914.html café]), it has been suggested to add a flag `-fshow-runtime-rep`. Without this flag enabled, the pretty printer will instantiate any `RuntimeRep` type parameters to `PtrRep Lifted`. This has the effect of changing {{{ ($) :: forall (r :: RuntimeRep) (a :: *) (b :: TYPE r). (a -> b) -> a -> b }}} to {{{ ($) :: (a -> b) -> a -> b }}} under the default GHCi settings. Note that `Levity` becomes `RuntimeRep` after #11471 is complete. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11549> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 20

[GHC] #10184: Coercible solver incomplete with recursive newtypes
by GHC 29 Jul '18

29 Jul '18

#10184: Coercible solver incomplete with recursive newtypes -------------------------------------+------------------------------------- Reporter: goldfire | Owner: Type: bug | Status: new Priority: low | Milestone: ⊥ Component: Compiler | Version: 7.10.1-rc3 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Blocked By: Test Case: | Related Tickets: Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- The following module fails to compile due to infinite recursion: {{{ import Data.Coerce newtype Bar a = Bar (Either a (Bar a)) newtype Age = MkAge Int x :: Bar Age x = coerce (Bar (Left (5 :: Int))) }}} It tries to coerce `Bar Int` to `Bar Age`. This is clearly doable, via tyconapp decomposition. But we don't find it. Before worrying too much about it, though, we'll wait for a real example. Do speak up if this problem is ruining your day! -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10184> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #14247: Fails to coerce between newtypes directly
by GHC 29 Jul '18

29 Jul '18

#14247: Fails to coerce between newtypes directly -------------------------------------+------------------------------------- Reporter: Iceland_jack | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.2.1 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: -------------------------------------+------------------------------------- {{{#!hs import Data.Coerce import Data.Type.Coercion newtype W a = W a newtype A = MkA (W A) a :: Coercion A A a = Coercion b :: Coercion a a' -> Coercion a (W a') b Coercion = Coercion c :: Coercion A (W A) c = b a }}} works just fine but the following fail: {{{#!hs -- • Couldn't match representation of type ‘A’ with that of ‘W A’ -- arising from a use of ‘Coercion’ -- • In the expression: Coercion -- In an equation for ‘d’: d = Coercion d :: Coercion A (W A) d = Coercion -- • Couldn't match representation of type ‘A’ with that of ‘W A’ -- arising from a use of ‘coerce’ -- • In the expression: coerce -- In an equation for ‘e’: e = coerce e :: A -> W A e = coerce }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14247> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #9936: Data.Fixed truncates 5.17 to 5.16
by GHC 28 Jul '18

28 Jul '18

#9936: Data.Fixed truncates 5.17 to 5.16 -------------------------------------+------------------------------------- Reporter: singpolyma | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: | Version: 7.6.3 libraries/base | Operating System: Unknown/Multiple Keywords: | Type of failure: Incorrect result Architecture: | at runtime Unknown/Multiple | Blocked By: Test Case: | Related Tickets: Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- `(realToFrac (5.17 :: Double) :: Centi) == 5.16` is true -- it should be false. The offender seems to be the assumption in `fromRational`: `fromRational r = withResolution (\res -> MkFixed (floor (r * (toRational res))))` Uses `floor` assuming that the underlying floating point value is stored purely and not as 0.999999 or similar -- switching it to `round` fixes this issue. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9936> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 13

[GHC] #10919: ghc: panic! (the 'impossible' happened) ... Dynamic linker not initialised
by GHC 28 Jul '18

28 Jul '18

#10919: ghc: panic! (the 'impossible' happened) ... Dynamic linker not initialised -------------------------------------+------------------------------------- Reporter: recursion- | Owner: ninja | Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.2 Keywords: panic | Operating System: Linux impossible linker initialized | Architecture: | Type of failure: Compile-time Unknown/Multiple | crash Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | -------------------------------------+------------------------------------- When trying to build the CUDA accelerate backend I get the following "complier panic": {{{ me@box:~/accelerate-cuda-0.15.0.0$ cabal build Building accelerate-cuda-0.15.0.0... Preprocessing library accelerate-cuda-0.15.0.0... Data/Array/Accelerate/CUDA/CodeGen/Base.hs:5:14: Warning: -XOverlappingInstances is deprecated: instead use per-instance pragmas OVERLAPPING/OVERLAPPABLE/OVERLAPS [ 7 of 33] Compiling Data.Array.Accelerate.CUDA.CodeGen.Type ( Data/Array/Accelerate/CUDA/CodeGen/Type.hs, dist/build/Data/Array/Accelerate/CUDA/CodeGen/Type.o ) <no location info>: ghc: panic! (the 'impossible' happened) (GHC version 7.10.2 for x86_64-unknown-linux): Dynamic linker not initialised Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug [ 9 of 33] Compiling Data.Array.Accelerate.CUDA.Debug ( Data/Array/Accelerate/CUDA/Debug.hs, dist/build/Data/Array/Accelerate/CUDA/Debug.o ) <no location info>: ghc: panic! (the 'impossible' happened) (GHC version 7.10.2 for x86_64-unknown-linux): Dynamic linker not initialised Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug [10 of 33] Compiling Data.Array.Accelerate.CUDA.Analysis.Launch ( Data/Array/Accelerate/CUDA/Analysis/Launch.hs, dist/build/Data/Array/Accelerate/CUDA/Analysis/Launch.o ) <no location info>: <command line>: can't load .so/.DLL for: /home/me/.cabal/lib/x86_64 -linux-ghc-7.10.2/cuda-0.6.7.0-EMlxEWvymgZGXnLSaVKypt/libHScuda-0.6.7.0 -EMlxEWvymgZGXnLSaVKypt-ghc7.10.2.so (libcudart.so.7.5: cannot open shared object file: No such file or directory) }}} This is the accelerate cuda package version 0.15.0.0: https://hackage.haskell.org/package/accelerate-cuda I changed the constraints from base==4.7.* to base==4.8.* to allow for compilation with GHC 7.10.2. After installing the dependencies through cabal and attempting to manually build the package manuall with cabal build I get the above mentioned compiler panics. GHC instructed me to report the issue as a bug (defect) so I did. I'd like to get this library to build with the newest version of GHC so please let me know how to proceed. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10919> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #12694: GHC HEAD no longer reports inaccessible code
by GHC 27 Jul '18

27 Jul '18

#12694: GHC HEAD no longer reports inaccessible code -------------------------------------+------------------------------------- Reporter: ezyang | Owner: Type: bug | Status: new 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 this A.hs: {{{ {-# LANGUAGE GADTs #-} module A where f :: Bool ~ Int => a -> b f x = x }}} I get: {{{ ezyang@sabre:~$ ghc-head --version The Glorious Glasgow Haskell Compilation System, version 8.1.20161010 ezyang@sabre:~$ ghc-head --make A.hs [1 of 1] Compiling A ( A.hs, A.o ) A.hs:4:1: warning: [-Woverlapping-patterns] Pattern match is redundant In an equation for ‘f’: f x = ... }}} In contrast: {{{ ezyang@sabre:~$ ghc-8.0 --make A.hs [1 of 1] Compiling A ( A.hs, A.o ) A.hs:3:6: error: • Couldn't match type ‘Bool’ with ‘Int’ Inaccessible code in the type signature for: f :: Bool ~ Int => a -> b • In the ambiguity check for ‘f’ To defer the ambiguity check to use sites, enable AllowAmbiguousTypes In the type signature: f :: Bool ~ Int => a -> b }}} Is this expected? I'd expect at least a warning! -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12694> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #13766: Confusing "reudundant pattern match" in 8.0, no warning at all in 8.2
by GHC 27 Jul '18

27 Jul '18

#13766: Confusing "reudundant pattern match" in 8.0, no warning at all in 8.2 -------------------------------------+------------------------------------- Reporter: edsko | 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: -------------------------------------+------------------------------------- I couldn't find an existing ticket about this, so I figured I'd file this even though it's probably an instance of a more general problem. Consider {{{#!hs {-# LANGUAGE TypeFamilies #-} {-# OPTIONS_GHC -Wall -Woverlapping-patterns #-} module T where class C a where c :: a -> a instance Int ~ Bool => C Int where c = id }}} yields {{{ T.hs:10:3: warning: [-Woverlapping-patterns] Pattern match is redundant In an equation for ‘c’: c = ... }}} which I suppose makes _some_ amount of sense but is highly confusing nonetheless :) Now in 8.2 I don't get this warning, but in fact I don't get any warning at all, which I'm not entirely sure is better. In the real code obviously the superclass constraint was far more complicated and it was nice of ghc to warn me (albeit in a roundabout way) that it was unsatisfiable. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13766> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #14185: Non-local bug reporting around levity polymorphism
by GHC 27 Jul '18

27 Jul '18

#14185: Non-local bug reporting around levity polymorphism -------------------------------------+------------------------------------- Reporter: goldfire | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.3 Keywords: | Operating System: Unknown/Multiple LevityPolymorphism, | TypeErrorMessages | Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- From comment:12:ticket:13105. In HEAD (or, at least, my version, on the `wip/rae` branch), this code {{{#!hs class Unbox (t :: *) (r :: TYPE k) | t -> r, r -> t where unbox :: t -> r box :: r -> t instance Unbox Int Int# where unbox (I# i) = i box i = I# i instance Unbox Char Char# where unbox (C# c) = c box c = C# c instance (Unbox a a', Unbox b b') => Unbox (a,b) (# a', b' #) where unbox (a,b) = (# unbox a, unbox b #) box (# a, b #) = (box a, box b) testInt :: Int testInt = box (unbox 1) testTup :: (Int, Char) testTup = box (unbox (1, 'a')) }}} fails with {{{ Bug.hs:27:11: error: • Couldn't match a lifted type with an unlifted type When matching types a' :: * Int# :: TYPE 'IntRep • In the expression: box (unbox 1) In an equation for ‘testInt’: testInt = box (unbox 1) | 27 | testInt = box (unbox 1) | ^^^^^^^^^^^^^ Bug.hs:27:16: error: • Couldn't match a lifted type with an unlifted type When matching types a' :: * Int# :: TYPE 'IntRep • In the first argument of ‘box’, namely ‘(unbox 1)’ In the expression: box (unbox 1) In an equation for ‘testInt’: testInt = box (unbox 1) | 27 | testInt = box (unbox 1) | ^^^^^^^ Bug.hs:42:11: error: • Couldn't match a lifted type with an unlifted type When matching types a' :: * Int# :: TYPE 'IntRep • In the expression: box (unbox (1, 'a')) In an equation for ‘testTup’: testTup = box (unbox (1, 'a')) | 42 | testTup = box (unbox (1, 'a')) | ^^^^^^^^^^^^^^^^^^^^ }}} I think it should succeed. Worse, when I comment out the `testTup` definition, the file succeeds... but note that two of the errors above are in `testInt`, which compiles fine on its own. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14185> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #13116: Allow Overloaded things in patterns
by GHC 26 Jul '18

26 Jul '18

#13116: Allow Overloaded things in patterns -------------------------------------+------------------------------------- Reporter: | Owner: facundo.dominguez | Type: feature | Status: new request | 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 ticket is motivated by the use of `OverloadedLabels` of the [http://hackage.haskell.org/package/choice-0.2.0 choice package]. One can use a type `Choice "block"` with values `Do #block` and `Don't #block` by using `OverloadedLabels`. However, in patterns this syntax is forbidden. {{{ case block of Do #block -> ... Don't #block -> ... }}} It would be nice if overloaded labels desugared to something sensible in patterns, perhaps {{{ case block of Do ((==) (fromLabel (proxy# :: Proxy# "foo")) -> True) -> ... Don't ((==) (fromLabel (proxy# :: Proxy# "foo")) -> True) -> ... }}} `OverloadedLists` and `OverloadedStrings` could benefit from a similar solution. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13116> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 6

[GHC] #12736: Calling a complex Haskell function (obtained via FFI wrapper function) from MSVC 64-bit C code (passed in as FunPtr) can leave SSE2 registers in the XMM6-XMM15 range modified
by GHC 24 Jul '18

24 Jul '18

#12736: Calling a complex Haskell function (obtained via FFI wrapper function) from MSVC 64-bit C code (passed in as FunPtr) can leave SSE2 registers in the XMM6-XMM15 range modified -------------------------------------+------------------------------------- Reporter: bavism | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.3 (FFI) | Keywords: | Operating System: Windows ffi,registers,sse2,clobber,xmm | Architecture: x86_64 | Type of failure: Incorrect result (amd64) | at runtime Test Case: | Blocked By: https://github.com/bavis-m/raycast | Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- According to the [https://msdn.microsoft.com/en-us/library/9z1stfyw.aspx MSDN], in the Microsoft x64 architecture function calls must preserve the SSE2 registers in the range XMM6-XMM15. The Haskell FFI can produce a function pointer via dynamic wrapper that, when called from MSVC x64 C code, does not preserve these registers, causing further floating-point operations in the C code to fail. I can reproduce this error in [https://github.com/bavis-m/raycast this project], which is a DOOM-style raycasting engine written in Haskell, that imports a C DLL with glue for rendering and window management. The Haskell executable generates a FunPtr to a frame update function using the dynamic import mechanism, and passes this to a long-lived C function that runs the update loop. Any time this update function is called from the C loop, subsequent floating point operations produce incorrect results (in this case, the next operations compute a view matrix for the OpenGL window). The output on every frame showing the view matrix should be: {{{ viewM: 0.003125 0.000000 0.000000 -1.000000 0.000000 0.004167 0.000000 -1.000000 0.000000 0.000000 1.000000 0.000000 0.000000 0.000000 0.000000 1.000000 }}} Running the raycaster with the Release version of the DLL causes the value of this matrix to be corrupted. There is a patch provided (stub.patch in the root folder) that turns the Haskell update function into an empty stub. This causes the program to work. When stepping through the assembly code with this patch applied, I can see in the function prologue where the XMM registers are saved. Without the patch, these registers are not saved. Running the Debug version does not show this error; the register allocation must be different. I have been attempting to create a much simpler test case to reveal this code-generation issue, however it has been difficult. Even seemingly trivial changes can cause the bug to not show up, it is clearly dependent on the register allocation used internally to produce the assembly code. Instructions for building the project are in the readme. (You will need the Haskell Stack Tool, and Visual Studio 15). -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12736> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4