April 2013 - ghc-tickets

[GHC] #7835: ghc --make: allow passign .cmm, .c and .hs files in one command line
by GHC 14 Apr '13

14 Apr '13

#7835: ghc --make: allow passign .cmm, .c and .hs files in one command line -----------------------------+---------------------------------------------- Reporter: slyfox | Owner: Type: feature request | Status: new Priority: normal | Component: Compiler Version: 7.6.2 | Keywords: Os: Unknown/Multiple | Architecture: Unknown/Multiple Failure: None/Unknown | Blockedby: Blocking: | Related: -----------------------------+---------------------------------------------- TL;DR: Allow users to make one run: {{{ $ ghc --make Test.hs TestPrim.hs test-prims.cmm -o t }}} instead of series: {{{ $ ghc -c test-prims.cmm $ ghc --make test-prims.o Test.hs TestPrim.hs }}} The buildable example is located at: https://github.com/trofi/cmm-example I'll just quote my email sent to ghc-devs@ a while ago: {{{ > Hi guys! > > Today I've tried to write a minimal [1] test case for ghc on ia64 > (it SIGSEGVs there). > > The test case consists of a .cmm and some .hs modues. > > I tried to build them in one run as: > $ ghc --make Test.hs TestPrim.hs test-prims.cmm -o t > ghc: cannot compile this file to desired target: test-prims.cmm > Usage: For basic information, try the `--help' option. > > Thus had to use > ${GHC} -c test-prims.cmm > ${GHC} --make test-prims.o Test.hs TestPrim.hs > > What do you think of allowing to use .c, .cmm and similar > files along with .hs files to make one-shot builds a bit easier? > > If noone is opposed I'll fill a request on bugtracker. > > Thanks! > > [1]: https://github.com/trofi/cmm-example }}} -- Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/7835> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #7829: make better/more robust loopbreaker choices
by GHC 13 Apr '13

13 Apr '13

#7829: make better/more robust loopbreaker choices -----------------------------+---------------------------------------------- Reporter: nfrisby | Owner: Type: task | Status: new Priority: normal | Component: Compiler Version: 7.6.2 | Keywords: Os: Unknown/Multiple | Architecture: Unknown/Multiple Failure: None/Unknown | Blockedby: Blocking: | Related: -----------------------------+---------------------------------------------- The choice of loopbreaker can severely influence downstream compilation. This task ticket is about making the choice more robust/better/"smarter". This ticket is also empty of concrete suggestions how to do so... think of it like a community TODO. One example feature of the current algorithm that seems a bit fragile is the use of two schemes for breaking ties depending on the max "depth" of 2. Peruse the code and its comments and Notes in !OccAnal if you're interested. This also ticket serves to document a small regression incurred by my commit af12cf66d1a416a135cb98b86717aba2cd247e1a. There's a 4% increase in allocation in nofib/rewrite as a result of my change altering the loopbreaker choice. The actual details aren't relevant, but here's the basic story in order to convey the delicacy of loopbreaker choice. My commit slightly reduces the calculated size of a function in a mutually recursive group, so that it comes in under the "never unfold limit" instead of over. This ultimately causes the looperbreaker chooser to break a tie in a different way (there's two "Plans"). The previous choice was more fortuitous: it enabled a beneficial inlining that "misses its chance" with the new choice of loopbreaker. I don't remember nor ever totally understood the details of this last part of the story. I don't have the cycles at the moment to wade into it -dverbose-core2core -ddump-inlings again. Apologies. If a brave soul is interested, you should be able to recover the more fortuitous loopbreaker choice by setting `CoreUnfolding.sizeExpr.isRealWorldId` to `const False`. -- Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/7829> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1

[GHC] #7828: RebindableSyntax and Arrow
by GHC 13 Apr '13

13 Apr '13

#7828: RebindableSyntax and Arrow --------------------------------------+------------------------------------- Reporter: AlessandroVermeulen | Owner: Type: bug | Status: new Priority: normal | Component: Compiler Version: 7.6.2 | Keywords: Os: Unknown/Multiple | Architecture: Unknown/Multiple Failure: GHC rejects valid program | Blockedby: Blocking: | Related: --------------------------------------+------------------------------------- When trying to add constraints to the types of the arrow primitives I get a type error. I think that doing such a thing should be possible and I've attached the code I used to test this. The errors I get when using the arrow notation for the function test are as follows: {{{ test :: Typeable a => R a a test = proc n -> returnA -< n }}} {{{ bug-arrow.hs:15:8: Could not deduce (Typeable c) arising from a use of `arr' from the context (Typeable a) bound by the type signature for test :: Typeable a => R a a at bug-arrow.hs:14:9-27 Possible fix: add (Typeable c) to the context of a type expected by the context: (b -> c) -> R b c or the type signature for test :: Typeable a => R a a In the expression: arr When checking that `arr' (needed by a syntactic construct) has the required type: forall b1 c1. (b1 -> c1) -> R b1 c1 arising from a proc expression at bug-arrow.hs:15:8-29 In the expression: proc n -> returnA -< n bug-arrow.hs:15:8: Could not deduce (Typeable c) arising from a use of `>>>' from the context (Typeable a) bound by the type signature for test :: Typeable a => R a a at bug-arrow.hs:14:9-27 Possible fix: add (Typeable c) to the context of a type expected by the context: R a1 b -> R b c -> R a1 c or the type signature for test :: Typeable a => R a a In the expression: (>>>) When checking that `(>>>)' (needed by a syntactic construct) has the required type: forall a2 b1 c1. R a2 b1 -> R b1 c1 -> R a2 c1 arising from a proc expression at bug-arrow.hs:15:8-29 In the expression: proc n -> returnA -< n bug-arrow.hs:15:8: Could not deduce (Typeable d) arising from a use of `first' from the context (Typeable a) bound by the type signature for test :: Typeable a => R a a at bug-arrow.hs:14:9-27 Possible fix: add (Typeable d) to the context of a type expected by the context: R b c -> R (b, d) (c, d) or the type signature for test :: Typeable a => R a a In the expression: first When checking that `first' (needed by a syntactic construct) has the required type: forall b1 c1 d1. R b1 c1 -> R (b1, d1) (c1, d1) arising from a proc expression at bug-arrow.hs:15:8-29 In the expression: proc n -> returnA -< n }}} When I replace the definition with the translated core code (minus type applications and scoped type variables) the code compiles: {{{ test :: Typeable a => R a a test = (>>>) (arr (\ (n_apd) -> n_apd)) ((>>>) (arr (\ (ds_dst) -> ds_dst)) (returnA) ) }}} -- Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/7828> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #7821: PPC cross-compile impossible happened
by GHC 13 Apr '13

13 Apr '13

#7821: PPC cross-compile impossible happened --------------------------------+------------------------------------------- Reporter: matts | Owner: Type: bug | Status: new Priority: normal | Component: Compiler Version: 7.7 | Keywords: Os: Linux | Architecture: Unknown/Multiple Failure: Building GHC failed | Blockedby: Blocking: | Related: --------------------------------+------------------------------------------- While trying to build a powerpc targetted GHC cross compiler from current GIT head version I saw the impossible happened message below: echo 'export LD_LIBRARY_PATH=":$LD_LIBRARY_PATH"' >> inplace/bin/genapply echo 'exec "$executablename" ${1+"$@"}' >> inplace/bin/genapply chmod +x inplace/bin/genapply "inplace/bin/ghc-stage1" -hisuf hi -osuf o -hcsuf hc -static -H64m -O0 -fasm -package-name ghc-prim-0.3.1.0 -hide-all-packages -i -ilibraries/ghc-prim/. -ilibraries /ghc-prim/dist-install/ build -ilibraries/ghc-prim/dist-install/build/autogen -Ilibraries/ghc-prim /dist-install/build -I libraries/ghc-prim/dist-install/build/autogen -Ilibraries/ghc-prim/. -optP-include -optPlibra ries/ghc-prim/dist-install/build/autogen/cabal_macros.h -package rts-1.0 -package-name ghc-prim -XHaskell98 -XCPP -XMagicHash -XForeignFunctionInterface -XUnliftedFFITypes -XUnboxedTuples -XEm ptyDataDecls -XNoImplicitPrelude -O -fasm -no-user-package-db -rtsopts -odir libraries/ghc -prim/dist-install/build -hidir libraries/ghc-prim/dist-install/build -stubdir libraries/ghc-pri m/dist-install/build -dynamic-too -c libraries/ghc-prim/./GHC/Classes.hs -o libraries/ghc-prim/ dist-install/build/GHC/Classes.o -dyno libraries/ghc-prim/dist- install/build/GHC/Classes.dyn_o ghc-stage1: panic! (the 'impossible' happened) (GHC version 7.7.20130407 for powerpc-e500v2-linux): allocateRegsAndSpill: Cannot read from uninitialized register %vI_nvZF Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug make[1]: *** [libraries/ghc-prim/dist-install/build/GHC/Classes.o] Error 1 make[1]: *** Deleting file `libraries/ghc-prim/dist- install/build/GHC/Classes.o' -- Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/7821> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #7809: Silly error message suggesting LiberalTypeSynonyms
by GHC 13 Apr '13

13 Apr '13

#7809: Silly error message suggesting LiberalTypeSynonyms -----------------------------+---------------------------------------------- Reporter: goldfire | Owner: Type: bug | Status: new Priority: normal | Component: Compiler Version: 7.7 | Keywords: Os: Unknown/Multiple | Architecture: Unknown/Multiple Failure: None/Unknown | Blockedby: Blocking: | Related: -----------------------------+---------------------------------------------- The following code (rightly) does not compile: {{{ {-# LANGUAGE RankNTypes, TypeFamilies, LiberalTypeSynonyms #-} type PolyId = (forall a. a -> a) type family F a foo :: F PolyId foo = undefined }}} The error message is {{{ Illegal polymorphic or qualified type: PolyId Perhaps you intended to use -XLiberalTypeSynonyms In the type signature for `foo': foo :: F PolyId }}} The code doesn't compile because we can't apply a type family to a polytype. The first line of the error message says this well enough, though more information might be helpful here. However, the second line of the error message is silly: I've turned on {{{LiberalTypeSynonyms}}}! -- Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/7809> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1

[GHC] #7808: data families and TH names do not mix well (e.g. cannot use TH deriving)
by GHC 13 Apr '13

13 Apr '13

#7808: data families and TH names do not mix well (e.g. cannot use TH deriving) -----------------------------+---------------------------------------------- Reporter: duncan | Owner: Type: feature request | Status: new Priority: normal | Component: Template Haskell Version: 7.6.2 | Keywords: Os: Unknown/Multiple | Architecture: Unknown/Multiple Failure: None/Unknown | Blockedby: Blocking: | Related: -----------------------------+---------------------------------------------- lots of Haskell libraries have TH functions like: {{{ deriveJSON :: Name -> Q [Dec] }}} (e.g. aeson package for deriving JSON conversion) They're used like {{{ data Foo = Foo This That $(deriveJSON ''Foo) }}} the `deriveJSON` reifies the Name and expects to find that the name refers to a data type, from which it finds the data constructors and generates the appropriate code. Now, this doesn't work when the data type we want to derive an instance for is defined using a data family: {{{ class FooClass a where data Foo a ... instance FooClass Bar where data Foo Bar = Bar This That $(deriveJSON ''Foo Bar) -- ??? }}} We'd like to make the type Foo Bar an instance of the class (e.g. To/FromJSON) but `deriveJSON` expects a Name, and there is no name for this data type, its "name" is `Foo Bar` but obviously that's not a `TH.Name`. Of course in general a name for a type application doesn't make sense, but for data family instances what other way do we have to refer to them? -- Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/7808> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1

[GHC] #7807: Parse error with "where" and file-ending comment
by GHC 13 Apr '13

13 Apr '13

#7807: Parse error with "where" and file-ending comment -----------------------------+---------------------------------------------- Reporter: goldfire | Owner: Type: bug | Status: new Priority: normal | Component: Compiler (Parser) Version: 7.7 | Keywords: Os: Unknown/Multiple | Architecture: Unknown/Multiple Failure: None/Unknown | Blockedby: Blocking: | Related: -----------------------------+---------------------------------------------- Consider the following file: {{{ {-# LANGUAGE TypeFamilies #-} type family F a type instance where F Int = Bool {- -} }}} If there is a file-ending newline, after the comment, the file parses. If there is not a file-ending newline, I get a parse error. I wrote the code dealing with {{{type instance where}}} notation, but this seems more related to parsing issues (which I know little about) than {{{type instance where}}}. Please correct me if I'm wrong. -- Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/7807> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1

[GHC] #7806: Dynamic linking and libffi
by GHC 13 Apr '13

13 Apr '13

#7806: Dynamic linking and libffi ----------------------------------+----------------------------------------- Reporter: kazu-yamamoto | Owner: Type: bug | Status: new Priority: normal | Component: Build System Version: 7.7 | Keywords: Os: Unknown/Multiple | Architecture: Unknown/Multiple Failure: Installing GHC failed | Blockedby: Blocking: | Related: ----------------------------------+----------------------------------------- After dynamic linking became the default, building/installing GHC on FreeBSD and Mac fails due to libffi. I don't see any problems on Linux. = FreeBSD without system libffi = GHC can be build: {{{ % make maintainer-clean % perl boot % ./configure --prefix=/ghc-head \ --with-iconv-includes=/usr/local/include \ --with-iconv-libraries=/usr/local/lib \ --with-gmp-includes=/usr/local/include \ --with-gmp-libraries=/usr/local/lib \ --with-gcc=/usr/local/bin/gcc47 \ --with-gcc-4.2=/usr/local/bin/gcc47 % gmake -j10 }}} But installation fails: {{{ % gmake install Installing library in /ghc-head/lib/ghc-7.7.20130323/haskell2010-1.1.1.0 "/ghc-head/lib/ghc-7.7.20130323/bin/ghc-pkg" --force --global-package-db "/ghc-head/lib/ghc-7.7.20130323/package.conf.d" update rts/package.conf.install Shared object "libffi.so.6" not found, required by "libHSrts- ghc7.7.20130323.so" gmake[1]: *** [install_packages] Error 1 gmake: *** [install] Error 2 }}} = FreeBSD with system libffi = The following procedure has no problems: {{{ % make maintainer-clean % perl boot % ./configure --prefix=/ghc-head \ --with-iconv-includes=/usr/local/include \ --with-iconv-libraries=/usr/local/lib \ --with-gmp-includes=/usr/local/include \ --with-gmp-libraries=/usr/local/lib \ --with-gcc=/usr/local/bin/gcc47 \ --with-gcc-4.2=/usr/local/bin/gcc47 --with-system-libffi \ --with-ffi-includes=/usr/local/include \ --with-ffi-libraries=/usr/local/lib % gmake -j10 % gmake install }}} = Mac without system libffi = GHC can be build: {{{ % make maintainer-clean % perl boot % ./configure --prefix=/ghc-head % make -j6 }}} But installation fails: {{{ % make install Installing library in /ghc-head/lib/ghc-7.7.20130323/haskell2010-1.1.1.0 "/ghc-head/lib/ghc-7.7.20130323/bin/ghc-pkg" --force --global-package-db "/ghc-head/lib/ghc-7.7.20130323/package.conf.d" update rts/package.conf.install dyld: Library not loaded: @loader_path/../terminfo-0.3.2.5/libHSterminfo-0.3.2.5-ghc7.7.20130323.dylib Referenced from: /ghc-head/lib/ghc-7.7.20130323/bin/ghc-pkg Reason: image not found make[1]: *** [install_packages] Trace/BPT trap: 5 make: *** [install] Error 2 }}} = Mac with system libffi = If I do: {{{ % make maintainer-clean % perl boot % ./configure --prefix=/ghc-head \ --with-system-libffi \ --with-ffi-includes=/usr/include/ffi \ --with-ffi-libraries=/usr/lib % make -j6 }}} hsc2hs cannot find necessary .h files. {{{ "inplace/bin/hsc2hs" '--cc=/usr/bin/gcc' '--ld=/usr/bin/gcc' --cross-safe --cflag=-m64 --cflag=-fno-stack-protector --cflag=-m64 --lflag=-m64 --cflag=-D__GLASGOW_HASKELL__=707 --cflag=-Dx86_64_HOST_ARCH=1 --cflag=-Ddarwin_HOST_OS=1 '--cflag=-m64' '--cflag=-fno-stack-protector' ' --cflag=-Icompiler/../rts/dist/build' '--cflag=-Icompiler/stage2' '-- cflag=-Icompiler/.' '--cflag=-Icompiler/parser' '--cflag=-Icompiler/utils' '--cflag=-DGHCI' '-- cflag=-I/Users/kazu/work/ghc/libraries/process/include' '-- cflag=-I/Users/kazu/work/ghc/libraries/directory/include' '-- cflag=-I/Users/kazu/work/ghc/libraries/unix/include' '-- cflag=-I/Users/kazu/work/ghc/libraries/time/include' '-- cflag=-I/Users/kazu/work/ghc/libraries/containers/include' '-- cflag=-I/Users/kazu/work/ghc/libraries/bytestring/include' '-- cflag=-I/Users/kazu/work/ghc/libraries/array/include' '-- cflag=-I/Users/kazu/work/ghc/libraries/base/include' '-- cflag=-I/Users/kazu/work/ghc/rts/dist/build' '-- cflag=-I/Users/kazu/work/ghc/includes' '-- cflag=-I/Users/kazu/work/ghc/includes/dist-derivedconstants/header' '-- lflag=-m64' '--lflag=-L/Users/kazu/work/ghc/libraries/transformers/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/template-haskell /dist-install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/hpc/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/hoopl/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/bin-package-db /dist-install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/binary /dist-install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/Cabal/Cabal /dist-install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/process /dist-install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/pretty /dist-install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/directory /dist-install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/unix/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/time/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/old-locale/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/filepath/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/containers/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/bytestring/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/deepseq/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/array/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/base/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/integer-gmp/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/libraries/ghc-prim/dist- install/build' '--lflag=-L/Users/kazu/work/ghc/rts/dist/build' '-- lflag=-ldl' '--lflag=-liconv' '--lflag=-lm' '--lflag=-ldl' '--lflag=-lffi' --cflag=-Icompiler/stage2/build/autogen --cflag=-include --cflag=compiler/stage2/build/autogen/cabal_macros.h compiler/ghci/LibFFI.hsc -o compiler/stage2/build/LibFFI.hs LibFFI.hsc:9:17: error: ffi.h: No such file or directory LibFFI.hsc: In function ‘main’: LibFFI.hsc:42: error: ‘ffi_cif’ undeclared (first use in this function) LibFFI.hsc:42: error: (Each undeclared identifier is reported only once LibFFI.hsc:42: error: for each function it appears in.) LibFFI.hsc:82: error: ‘ffi_status’ undeclared (first use in this function) LibFFI.hsc:82: error: expected expression before ‘int’ LibFFI.hsc:82: error: expected ‘)’ before numeric constant LibFFI.hsc:82: error: expected ‘)’ before numeric constant LibFFI.hsc:83: error: ‘ffi_abi’ undeclared (first use in this function) LibFFI.hsc:83: error: expected expression before ‘int’ LibFFI.hsc:83: error: expected ‘)’ before numeric constant LibFFI.hsc:83: error: expected ‘)’ before numeric constant LibFFI.hsc:99: error: ‘FFI_OK’ undeclared (first use in this function) LibFFI.hsc:106: error: ‘FFI_DEFAULT_ABI’ undeclared (first use in this function) compiling compiler/stage2/build/LibFFI_hsc_make.c failed (exit code 1) command was: /usr/bin/gcc -c compiler/stage2/build/LibFFI_hsc_make.c -o compiler/stage2/build/LibFFI_hsc_make.o -m64 -fno-stack-protector -I/Users/kazu/work/ghc/includes -m64 -fno-stack-protector -m64 -D__GLASGOW_HASKELL__=707 -Dx86_64_HOST_ARCH=1 -Ddarwin_HOST_OS=1 -m64 -fno-stack-protector -Icompiler/../rts/dist/build -Icompiler/stage2 -Icompiler/. -Icompiler/parser -Icompiler/utils -DGHCI -I/Users/kazu/work/ghc/libraries/process/include -I/Users/kazu/work/ghc/libraries/directory/include -I/Users/kazu/work/ghc/libraries/unix/include -I/Users/kazu/work/ghc/libraries/time/include -I/Users/kazu/work/ghc/libraries/containers/include -I/Users/kazu/work/ghc/libraries/bytestring/include -I/Users/kazu/work/ghc/libraries/array/include -I/Users/kazu/work/ghc/libraries/base/include -I/Users/kazu/work/ghc/rts/dist/build -I/Users/kazu/work/ghc/includes -I/Users/kazu/work/ghc/includes/dist-derivedconstants/header -Icompiler/stage2/build/autogen -include compiler/stage2/build/autogen/cabal_macros.h -I/Users/kazu/work/ghc/inplace/lib/include/ make[1]: *** [compiler/stage2/build/LibFFI.hs] Error 1 make: *** [all] Error 2 }}} -- Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/7806> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 15

[GHC] #7800: Error message when deriving Typeable without PolyKinds
by GHC 13 Apr '13

13 Apr '13

#7800: Error message when deriving Typeable without PolyKinds ----------------------------------------------+----------------------------- Reporter: monoidal | Owner: Type: bug | Status: new Priority: normal | Component: Compiler Version: 7.7 | Keywords: Os: Unknown/Multiple | Architecture: Unknown/Multiple Failure: Incorrect warning at compile-time | Blockedby: Blocking: | Related: ----------------------------------------------+----------------------------- Module M: {{{ {-# LANGUAGE PolyKinds #-} module M where data A a }}} Module N: {{{ {-# LANGUAGE StandaloneDeriving, DeriveDataTypeable #-} module N where import M import Data.Typeable deriving instance Typeable A }}} We get a confusing error message: {{{ N.hs:7:1: Derived typeable instance must be of form (Typeable A) In the stand-alone deriving instance for ‛Typeable A’ }}} The error disappears when N is compiled with -XPolyKinds. The reason is: without -XPolyKinds, A defaults to A *. I think it's simplest to suggest the flag in the error message. -- Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/7800> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 2

[GHC] #7797: re-enable the defun RULE from a SPECIALISE instance pragma
by GHC 13 Apr '13

13 Apr '13

#7797: re-enable the defun RULE from a SPECIALISE instance pragma -----------------------------+---------------------------------------------- Reporter: nfrisby | Owner: Type: bug | Status: new Priority: normal | Component: Compiler Version: 7.6.2 | Keywords: Os: Unknown/Multiple | Architecture: Unknown/Multiple Failure: None/Unknown | Blockedby: Blocking: | Related: -----------------------------+---------------------------------------------- As of commit [https://github.com/ghc/ghc/commit/51d89a55c3 51d89a55c3], `SPECIALISE instance` pragmas do not result in a RULE for the dictionary function. For example, consider this `Eq` instance for `List`. {{{ module M where data List a = Nil | Cons a (List a) instance (Eq a) => Eq (List a) where {-# SPECIALISE instance Eq (List Char) #-} Nil == Nil = True (Cons x xs) == (Cons y ys) = x == y && xs == ys _xs == _ys = False }}} With ghc-7.4.2, we get: {{{ ==================== Tidy Core rules ==================== "SPEC $c/=" [ALWAYS] forall ($dEq :: GHC.Classes.Eq GHC.Types.Char). M.$fEqList_$c/=1 @ GHC.Types.Char $dEq = M.$fEqList_$c/= "SPEC $c==" [ALWAYS] forall ($dEq :: GHC.Classes.Eq GHC.Types.Char). M.$fEqList_$c==1 @ GHC.Types.Char $dEq = M.$fEqList_$c== "SPEC M.$fEqList" [ALWAYS] forall ($dEq :: GHC.Classes.Eq GHC.Types.Char). M.$fEqList @ GHC.Types.Char $dEq = M.$fEqList_$fEqList }}} Note the last rule: it specializes the normal defun `M.$fEqList` when applied at type `Char`. With anything after 7.4.2 -- eg if you download the binary sources for 7.4.2 and make [https://github.com/ghc/ghc/commit/51d89a55c3#L0L779 the change at line 779] from that commit --- you instead get this: {{{ ==================== Tidy Core rules ==================== "SPEC $c/=" [ALWAYS] forall ($dEq :: GHC.Classes.Eq GHC.Types.Char). M.$fEqList_$c/=1 @ GHC.Types.Char $dEq = M.$fEqList_$c/= "SPEC $c==" [ALWAYS] forall ($dEq :: GHC.Classes.Eq GHC.Types.Char). M.$fEqList_$c==1 @ GHC.Types.Char $dEq = M.$fEqList_$c== }}} (Actually, after some patch the /= RULE disappears too, but I don't know which/why.) If the dictionary is used at the relevant type in the same module, the specializer will automatically create the omitted rule. That will not, however, currently happen across module boundaries. In my contrived example, omitting this defun specialization increases runtime by a factor of 2 at -O1. {{{ {-# LANGUAGE ExistentialQuantification #-} module Main where import M data Box = forall a. Eq a => Box a a box = Box (go 10000000) (go 10000000) where go :: Int -> List Char go 0 = Nil go n = Cons 'c' $ go (n - 1) {-# NOINLINE box #-} main = print $ case box of Box l r -> l == r }}} (-O2 squashes the runtime difference; I haven't investigated in detail.) -- Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/7797> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3