- ghc-tickets - Haskell.org

[GHC] #16357: Add `oneShot` to the implementation of foldlM
by GHC 25 Feb '19

25 Feb '19

#16357: Add `oneShot` to the implementation of foldlM -------------------------------------+------------------------------------- Reporter: autotaker | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: | Version: 8.9 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: -------------------------------------+------------------------------------- The current (473632d7671619ee08a2a0025aa22bd4f79eca2d) implementation of `Data.Foldable.foldlM` is the like this {{{#!hs foldlM :: (Foldable t, Monad m) => (b -> a -> m b) -> b -> t a -> m b foldlM f z0 xs = foldr c return xs z0 -- See Note [List fusion and continuations in 'c'] where c x k z = f z x >>= k {-# INLINE c #-} }}} It generates an inefficient core for the following example. {{{#!hs f :: Int -> IO Int f = foldlM (\a b -> pure $! (a + b)) 0 (filter even [1..n]) }}} Generated core: {{{#!hs -- RHS size: {terms: 48, types: 22, coercions: 12, joins: 0/1} Main.$wf [InlPrag=NOUSERINLINE[2]] :: GHC.Prim.Int# -> GHC.Prim.State# GHC.Prim.RealWorld -> (# GHC.Prim.State# GHC.Prim.RealWorld, Int #) [GblId, Arity=2, Caf=NoCafRefs, Str=<L,U><L,U>, Unf=Unf{Src=<vanilla>, TopLvl=True, Value=True, ConLike=True, WorkFree=True, Expandable=True, Guidance=IF_ARGS [0 0] 216 30}] Main.$wf = \ (ww_s6TZ :: GHC.Prim.Int#) (w_s6TW :: GHC.Prim.State# GHC.Prim.RealWorld) -> case GHC.Prim.># 1# ww_s6TZ of { __DEFAULT -> letrec { go_a5un [Occ=LoopBreaker] :: GHC.Prim.Int# -> Int -> IO Int [LclId, Arity=1, Str=<L,U>, Unf=OtherCon []] go_a5un = \ (x_a5uo :: GHC.Prim.Int#) -> case GHC.Prim.remInt# x_a5uo 2# of { __DEFAULT -> case GHC.Prim.==# x_a5uo ww_s6TZ of { __DEFAULT -> go_a5un (GHC.Prim.+# x_a5uo 1#); 1# -> (GHC.Base.$fApplicativeIO4 @ Int) `cast` (<Int>_R ->_R Sym (GHC.Types.N:IO[0] <Int>_R) :: (Int -> GHC.Prim.State# GHC.Prim.RealWorld -> (# GHC.Prim.State# GHC.Prim.RealWorld, Int #)) ~R# (Int -> IO Int)) }; 0# -> Main.main_c @ Int (GHC.Types.I# x_a5uo) (case GHC.Prim.==# x_a5uo ww_s6TZ of { __DEFAULT -> go_a5un (GHC.Prim.+# x_a5uo 1#); 1# -> (GHC.Base.$fApplicativeIO4 @ Int) `cast` (<Int>_R ->_R Sym (GHC.Types.N:IO[0] <Int>_R) :: (Int -> GHC.Prim.State# GHC.Prim.RealWorld -> (# GHC.Prim.State# GHC.Prim.RealWorld, Int #)) ~R# (Int -> IO Int)) }) }; } in ((go_a5un 1# Main.main4) `cast` (GHC.Types.N:IO[0] <Int>_R :: IO Int ~R# (GHC.Prim.State# GHC.Prim.RealWorld -> (# GHC.Prim.State# GHC.Prim.RealWorld, Int #)))) w_s6TW; 1# -> (# w_s6TW, Main.main4 #) } }}} It seems that the main loop `go_a5un` is not eta-expanded. I think problem is that `oneShot` is missing in the definition of `foldlM`. When I changed the definition of `foldlM` as follows, {{{#!hs import GHC.Exts(oneShot) foldlM :: (Foldable t, Monad m) => (b -> a -> m b) -> b -> t a -> m b foldlM f z0 xs = foldr c return xs z0 -- See Note [List fusion and continuations in 'c'] where c x k = oneShot (\z -> f z x >>= k) {-# INLINE c #-} }}} Then, the main loop of the `wf` is eta-expaned as expected. {{{#!hs -- RHS size: {terms: 64, types: 46, coercions: 0, joins: 1/1} Main.$wf [InlPrag=NOUSERINLINE[2]] :: GHC.Prim.Int# -> GHC.Prim.State# GHC.Prim.RealWorld -> (# GHC.Prim.State# GHC.Prim.RealWorld, Int #) [GblId, Arity=2, Caf=NoCafRefs, Str=<L,U><L,U>, Unf=Unf{Src=<vanilla>, TopLvl=True, Value=True, ConLike=True, WorkFree=True, Expandable=True, Guidance=IF_ARGS [0 0] 136 30}] Main.$wf = \ (ww_s6Xc :: GHC.Prim.Int#) (w_s6X9 :: GHC.Prim.State# GHC.Prim.RealWorld) -> case GHC.Prim.># 1# ww_s6Xc of { __DEFAULT -> joinrec { go_s6WG [Occ=LoopBreaker] :: GHC.Prim.Int# -> Int -> GHC.Prim.State# GHC.Prim.RealWorld -> (# GHC.Prim.State# GHC.Prim.RealWorld, Int #) [LclId[JoinId(3)], Arity=3, Str=<L,U><L,U(U)><L,U>, Unf=OtherCon []] go_s6WG (x_a5xy :: GHC.Prim.Int#) (eta_B2 :: Int) (eta1_Xz :: GHC.Prim.State# GHC.Prim.RealWorld) = case GHC.Prim.remInt# x_a5xy 2# of { __DEFAULT -> case GHC.Prim.==# x_a5xy ww_s6Xc of { __DEFAULT -> jump go_s6WG (GHC.Prim.+# x_a5xy 1#) eta_B2 eta1_Xz; 1# -> (# eta1_Xz, eta_B2 #) }; 0# -> case eta_B2 of { GHC.Types.I# x1_a5t8 -> case GHC.Prim.==# x_a5xy ww_s6Xc of { __DEFAULT -> jump go_s6WG (GHC.Prim.+# x_a5xy 1#) (GHC.Types.I# (GHC.Prim.+# x1_a5t8 x_a5xy)) eta1_Xz; 1# -> (# eta1_Xz, GHC.Types.I# (GHC.Prim.+# x1_a5t8 x_a5xy) #) } } }; } in jump go_s6WG 1# Main.main4 w_s6X9; 1# -> (# w_s6X9, Main.main4 #) } }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/16357> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #16130: GHC Panic on OS X: Data.Binary.Get.runGet: Invalid magic number "INPUT(-l"
by GHC 25 Feb '19

25 Feb '19

#16130: GHC Panic on OS X: Data.Binary.Get.runGet: Invalid magic number "INPUT(-l" ----------------------------------------+--------------------------------- Reporter: basvandijk | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.6.3 Keywords: | Operating System: MacOS X Architecture: Unknown/Multiple | Type of failure: None/Unknown Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: ----------------------------------------+--------------------------------- When building the latest `opencv-extra` library on OS X I'm getting a GHC panic. To reproduce git clone https://github.com/LumiGuide/haskell-opencv, make sure Nix is installed and execute: {{{ > nix-build -A haskellPackages.opencv-extra ... [11 of 11] Compiling OpenCV.Extra ( src/OpenCV/Extra.hs, dist/build/OpenCV/Extra.p_o ) ghc: panic! (the 'impossible' happened) (GHC version 8.6.3 for x86_64-apple-darwin): Data.Binary.Get.runGet at position 8: Invalid magic number "INPUT(-l" CallStack (from HasCallStack): error, called at libraries/binary/src/Data/Binary/Get.hs:351:5 in binary-0.8.6.0:Data.Binary.Get }}} The library builds successfully on Linux. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/16130> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 11

[GHC] #16100: T11627a fails sometimes on CI
by GHC 24 Feb '19

24 Feb '19

#16100: T11627a fails sometimes on CI -------------------------------------+------------------------------------- Reporter: mpickering | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.6.3 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: -------------------------------------+------------------------------------- See this build - https://gitlab.haskell.org/mpickering/ghc/-/jobs/6459 Causes this panic {{{ 81384 cd "profiling/should_run/T11627b.run" && ./T11627b +RTS -hd -RTS +RTS -i0 -RTS 81385 Wrong exit code for T11627a(prof_hc_hb)(expected 0 , actual 134 ) 81386 Stdout ( T11627a ): 81387 456574 81388 Stderr ( T11627a ): 81389 T11627a: internal error: LDV_recordDead: Failed to find counter for closure 0x42000bf558 81390 (GHC version 8.7.20181227 for x86_64_unknown_linux) 81391 Please report this as a GHC bug: https://www.haskell.org/ghc/reportabug }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/16100> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #16185: Add an AnonArgFlag to FunTy
by GHC 24 Feb '19

24 Feb '19

#16185: Add an AnonArgFlag to FunTy -------------------------------------+------------------------------------- Reporter: simonpj | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.6.3 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: -------------------------------------+------------------------------------- In Trac #15952 Richard and I agreed that it makes sense to add a flag to `FunTy` to distinguish `(->)` from `(=>)`. This ticket tracks progress. See [ticket:15952#comment:7] and [ticket:15952#comment:8]. Main changes: * Define `AnonArgFlag` in `Var`, alongside `ArgFlag` {{{ data AnonArgFlag = VisArg -- Used for (->): an ordinary non-dependent arrow -- The argument is visible in source code | InvisArg -- Used for (=>): a non-dependent predicate arrow -- The argument is invisible in source code }}} * `Type` looks like this {{{ data Type = ... | FFunTy -- ^ t1 -> t2 Very common, so an important special case -- FFunTy for "full function type"; see pattern synonym for FunTy { ft_af :: AnonArgFlag -- Is this (->) or (=>)? , ft_arg :: Type -- Argument type , ft_res :: Type } -- Resuult type }}} * Add a uni-directional pattern synonym for the old `FunTy` {{{ pattern FunTy :: Type -> Type -> Type pattern FunTy arg res <- FFunTy { ft_arg = arg, ft_res = res } }}} Everything else follows routinely. A huge win is that `isPredTy` vanishes almost completely. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/16185> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 12

[GHC] #16361: Non-deterministic hs_try_putmvar00 failure on CI
by GHC 24 Feb '19

24 Feb '19

#16361: Non-deterministic hs_try_putmvar00 failure on CI -------------------------------------+------------------------------------- Reporter: int-index | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Test Suite | Version: 8.9 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: -------------------------------------+------------------------------------- {{{ Wrong exit code for hs_try_putmvar003(threaded1)(expected 0 , actual 139 ) *** unexpected failure for hs_try_putmvar003(threaded1) }}} A quick search suggests that 139 might mean a segfault (SIGSEGV). This is a non-deterministic CI failure that happened to me at least two times (Feb 12 and Feb 24), and forced me to restart the build job to get a green build status. Both times it was the `validate-x86_64-linux-fedora27` job, but on different runners (`ghc-ci-gce-x86_64-1` and `maurer.smart- cactus.org`). Note that another failure that seems to happen only on Fedora is #16350 – could it be a problem with the Fedora image? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/16361> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #16360: GHC fails when GHC_PACKAGE_PATH contains trailing slash
by GHC 23 Feb '19

23 Feb '19

#16360: GHC fails when GHC_PACKAGE_PATH contains trailing slash -------------------------------------+------------------------------------- Reporter: nh2 | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.6.3 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: GHC doesn't work Unknown/Multiple | at all Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- When the directory given in `GHC_PACKAGE_PATH` contains a trailing slash, GHC stops working completely. For example, this works: {{{ env GHC_PACKAGE_PATH=/raid/src/ghc/ghc-atomic- writes/_build/stage1/lib/package.conf.d /raid/src/ghc/ghc-atomic- writes/_build/stage1/bin/ghc "--make" "-outputdir" "example-programs- build/" "example-programs/Hello.hs" "-o" "example-programs-build/haskell- hello" }}} but adding a trailing `/` like `GHC_PACKAGE_PATH=/path/to/package.conf.d/` fails: {{{ env GHC_PACKAGE_PATH=/raid/src/ghc/ghc-atomic- writes/_build/stage1/lib/package.conf.d/ /raid/src/ghc/ghc-atomic- writes/_build/stage1/bin/ghc "--make" "-outputdir" "example-programs- build/" "example-programs/Hello.hs" "-o" "example-programs-build/haskell- hello" }}} {{{ [1 of 1] Compiling Main ( example-programs/Hello.hs, example- programs-build/Main.o ) [Prelude changed] example-programs/Hello.hs:1:1: error: Could not find module ‘Prelude’ There are files missing in the ‘base-4.12.0.0’ package, try running 'ghc-pkg check'. Use -v (or `:set -v` in ghci) to see a list of the files searched for. | 1 | main :: IO () | ^ }}} This is surprising; most programs accept trailing slashes where directories are expected, and shells ofthen TAB-complete directories this way. It wastes developer time when they don't notice. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/16360> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 2

[GHC] #15029: haddock parsing fails with valid input
by GHC 23 Feb '19

23 Feb '19

#15029: haddock parsing fails with valid input -------------------------------------+------------------------------------- Reporter: lerkok | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.6.1 Component: Compiler | Version: 8.2.2 (Parser) | Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: Other Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Reported here: https://github.com/haskell/haddock/issues/798 This is really a GHC parsing issue, even though it only shows up in the haddock "route." -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15029> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1

[GHC] #16358: Confusing ghci error message
by GHC 23 Feb '19

23 Feb '19

#16358: Confusing ghci error message -------------------------------------+------------------------------------- Reporter: lerkok | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.6.3 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: -------------------------------------+------------------------------------- {{{ $ ghci GHCi, version 8.6.3: http://www.haskell.org/ghc/ :? for help Prelude> import qualified Data.Set as S Prelude S> S.foo <interactive>:2:1: error: Not in scope: ‘S.foo’ No module named ‘S’ is imported. }}} I'm curious why `ghci` says `No module named 'S' is imported`. I found it very confusing. I'd have expected it to say one of three things: 1. `Module Data.Set referenced via S does not export foo` 2. `Constructor S not in scope` 3. `Variable foo not in scope` The first one would be the most informative, but I can see that the second and the third messages might be quite reasonable as well due to the interpretation of dot as composition. But the message `No module S is imported` is quite confusing given I just issued `import qualified Data.Set as S`. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/16358> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 2

[GHC] #16353: ./validate --hadrian doesn't quite work on Windows
by GHC 22 Feb '19

22 Feb '19

#16353: ./validate --hadrian doesn't quite work on Windows -------------------------------------+------------------------------------- Reporter: alpmestan | Owner: (none) Type: bug | Status: new Priority: high | Milestone: Component: Build System | Version: 8.7 (Hadrian) | Keywords: | Operating System: Windows Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- On my Windows machine, using an msys2 shell as advised by the Windows- specific pages of GHC's trac, I managed to get `./validate --hadrian` (from [https://gitlab.haskell.org/ghc/ghc/merge_requests/326 this MR]) to go pretty far. I did manage to make it produce a binary distribution after some serious wrangling, to have it installed somewhere nearby and it ends up choking on [https://gitlab.haskell.org/alp/ghc/blob/wip/alp/hadrian- validate/validate#L258 this command] which tries to configure an `xhtml` build using `Setup.hs` scripts. When I looked at the verbose output, I ended up seeing that Cabal calls `findProgramVersion` for the ghc that I pass using `--with-ghc=...`, with `--numeric-version` as the only argument. And this ends up returning `""` -- which is what's returned when exceptions are thrown apparently, judging from [https://github.com/haskell/cabal/blob/c2cf30cc50109f2ffa30fd7affe9c22e1adf9… findProgramVersion's implementation]. Now, where this gets "interesting" is that I've tried to run `./Setup configure --with-ghc=X` for different values of `X`... - (1) `./Setup configure --with- ghc=/path/to/ghc/tree/_validatebuild/stage1/bin/ghc` works - (2) `./Setup configure --with- ghc=/path/to/ghc/tree/_validatebuild/bindist/ghc-8.6.20190220/bin/ghc` works - (3) `./Setup configure --with-ghc=/path/to/ghc/tree/bindisttest/install dir/bin/ghc` **doesn't work** (3 spaces between `install` and `dir`) - (4) `./Setup configure --with-ghc=/path/to/ghc/tree/a/b/bin/ghc` **doesn't work** (1) is our stage 2 GHC executable, the real one, produced right at this path. (2) is a copy of (1) that we put under this `bindist/ghc-X.Y.Z/` dir, which we end up putting wholesale in an archive. (3) is where `./validate` (whether with `--hadrian` or not) installs binary distributions. The install process goes fine, I can even use the GHC in question and evaluate `.../bin/ghc -e 'Data.Text.IO.putStrLn (Data.Text.pack "hello")'` without any problem, or get the expected output with `.../bin/ghc --numeric-version`. But the `configure` step just won't have it. I initially thought this was due to the directory with a space in it, one way or another, so I decided to try and install the bindist manually in a place with no spaces in the directory names, (4). (4) doesn't work either though. The main difference between 1) 2) on one hand, and 3) 4) on the other, is that in the first two cases I'm passing the path to an actual stage 2 GHC executable, while in the case of 3) and 4) I'm pointing to wrapper scripts, with the actual executables being under `$topdir/lib/bin/`. I'm however not sure of what's going on, it seems to be like this should work, any help is appreciated. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/16353> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #16343: Changing an irrelevant constraint leads to inlining failure
by GHC 22 Feb '19

22 Feb '19

#16343: Changing an irrelevant constraint leads to inlining failure -------------------------------------+------------------------------------- Reporter: mpickering | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.6.3 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 is an example Andres sent me but I can't work out precisely what is going on. {{{ {-# LANGUAGE AllowAmbiguousTypes #-} {-# LANGUAGE ConstraintKinds #-} {-# LANGUAGE DataKinds #-} {-# LANGUAGE FlexibleContexts #-} {-# LANGUAGE FlexibleInstances #-} {-# LANGUAGE GADTs #-} {-# LANGUAGE MultiParamTypeClasses #-} {-# LANGUAGE PolyKinds #-} {-# LANGUAGE QuantifiedConstraints #-} {-# LANGUAGE RankNTypes #-} {-# LANGUAGE ScopedTypeVariables #-} {-# LANGUAGE TemplateHaskell #-} {-# LANGUAGE TypeApplications #-} {-# LANGUAGE TypeOperators #-} module Tests where import Data.Kind import Data.Maybe (maybeToList) data Product (f :: a -> Type) (xs :: [a]) :: Type where Nil :: Product f '[] (:*) :: f x -> !(Product f xs) -> Product f (x : xs) infixr 5 :* class Top a instance Top a class (forall d . (forall x . c x => d x) => All d xs) => All (c :: a -> Constraint) (xs :: [a]) where ccataList :: f '[] -> (forall y ys . (c y, All c ys) => f ys -> f (y : ys)) -> f xs instance All c '[] where ccataList nil _cons = nil {-# INLINE ccataList #-} instance (c x, All c xs) => All c (x : xs) where ccataList nil cons = cons (ccataList @_ @c nil cons) {-# INLINE ccataList #-} newtype (f ~> g) x = Lambda { apply :: f x -> g x } cpureProduct :: forall c xs f . All c xs => (forall x . c x => f x) -> Product f xs cpureProduct p = ccataList @_ @c Nil (p :*) {-# INLINE cpureProduct #-} newtype Apply f g x = MkApply { unApply :: Product (f ~> g) x -> Product f x -> Product g x } capplyProduct :: forall c xs f g . All c xs => Product (f ~> g) xs -> Product f xs -> Product g xs capplyProduct = unApply $ ccataList @_ @c (MkApply (\ Nil Nil -> Nil)) (\ rec -> MkApply (\ (f :* fs) (x :* xs) -> apply f x :* unApply rec fs xs)) {-# INLINE capplyProduct #-} -- The addition of the constraint makes the test pass. mapProduct :: forall c xs f g . ({- All Top xs, -} All c xs) => (forall x . c x => f x -> g x) -> Product f xs -> Product g xs mapProduct p q = capplyProduct @Top (cpureProduct @c (Lambda p)) q {-# INLINE mapProduct #-} test :: Product [] [ (), (), () ] test = mapProduct @Top maybeToList (cpureProduct @Top Nothing) reference :: Product [] [ (), (), () ] reference = [] :* [] :* [] :* Nil }}} `test` should simplify to something close to `reference. There are a few ways to make it do so. 1. Uncomment the constraint in `mapProduct` 2. Change `Top` to `c` in `capplyProduct @Top`. From looking at `-ddump-inlinings`, it seems that things start to go wrong when `$cccataList` is inlined. In the good case, it looks like this. {{{ Considering inlining: $cccataList_a1Bi arg infos [ValueArg, ValueArg, ValueArg, ValueArg, NonTrivArg, NonTrivArg] interesting continuation RhsCtxt some_benefit True is exp: True is work-free: True guidance ALWAYS_IF(arity=4,unsat_ok=False,boring_ok=False) ANSWER = YES }}} In the bad case it looks like {{{ Considering inlining: $cccataList_a1Bv arg infos [ValueArg, ValueArg, ValueArg, ValueArg] interesting continuation RuleArgCtxt some_benefit True is exp: True is work-free: True guidance ALWAYS_IF(arity=4,unsat_ok=False,boring_ok=False) ANSWER = YES }}} Notice the continuation and arg infos are different for the two cases. The first case seems to be eta expanded? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/16343> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1