September 2014 - ghc-tickets

Re: [GHC] #2273: inlining defeats seq
by GHC 23 Sep '14

23 Sep '14

#2273: inlining defeats seq -------------------------------------+------------------------------------- Reporter: igloo | Owner: Type: bug | Status: new Priority: lowest | Milestone: 7.10.1 Component: Compiler | Version: 6.9 Resolution: | Keywords: Operating System: | Architecture: Unknown/Multiple Unknown/Multiple | Difficulty: Unknown Type of failure: | Blocked By: None/Unknown | Related Tickets: Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- Comment (by thomie): In [changeset:"970816ac0028f2f42ac4140d29e2f0dfe0e9af3e/ghc"] (Simon Marlow, September 2009): {{{ #!CommitTicketReference repository="ghc" revision="970816ac0028f2f42ac4140d29e2f0dfe0e9af3e" }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/2273#comment:24> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #8026: DatatypeContexts should be fixed, not deprecated
by GHC 23 Sep '14

23 Sep '14

#8026: DatatypeContexts should be fixed, not deprecated -------------------------------------+------------------------------------- Reporter: gidyn | Owner: Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.6.3 Component: Compiler | Keywords: Resolution: | Architecture: Unknown/Multiple Operating System: | Difficulty: Unknown Unknown/Multiple | Blocked By: Type of failure: | Related Tickets: None/Unknown | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- Comment (by gidyn): I think we actually have 2 implementation proposals here: - tokyo's rules add type constraints, whenever the data type appears in an explicit type signature or when the constructor is used - use regular type inference The second implementation would give different results for `c` and `d`: {{{ c :: Proxy (HasEq a) -> () -- what constraints are used here??? c _ = () }}} No type constraint is added, because none is inferred by the definition of `c`. {{{ data Dynamic = forall a. Typeable a => MkDyn a d :: Dynamic -> Maybe Bool d (MkDyn y) | Just (MkHasEq z) <- cast y = Just (z == z) -- where does the Eq constraint come from?? d _ = Nothing }}} `DatatypeContexts` wouldn't have any effect in this case. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8026#comment:20> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #8026: DatatypeContexts should be fixed, not deprecated
by GHC 22 Sep '14

22 Sep '14

#8026: DatatypeContexts should be fixed, not deprecated -------------------------------------+------------------------------------- Reporter: gidyn | Owner: Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.6.3 Component: Compiler | Keywords: Resolution: | Architecture: Unknown/Multiple Operating System: | Difficulty: Unknown Unknown/Multiple | Blocked By: Type of failure: | Related Tickets: None/Unknown | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- Comment (by yokto): Ok I'm not an expert in type theory or anything but let me try to explain based on your example. I don't know how this relates to type theory or if it is sound but it covers all the cases I could think of. {{{data Eq a => HasEq a = MkHasEq a}}} == Deriving Types == How can you get a value of type {{{HasEq a}}}. 1. Using a Constructor (also in a pattern): Use the same trick as with GADTs? Just make the type signature of the constructor {{{(Eq a) => a -> HasEq a}}} 2. Explicit type signature {{{(undefined :: HasEq a)}}}: Add the constraint {{{(Eq a)}}} to all explicit type signatures containing HasEq a. If you have more variables {{{HasEq (Either a b)}}} use the inner most forall in which one of the variables is bound {{{ (forall a. HasEq (Either a b) -> c) -> c -> b }}} becomes {{{ (forall a. (Eq (Either a b)) => HasEq (Either a b) -> c) -> c -> b }}} 3. Functions in modules without this extension. Maybe it's possible to add the contexts when the functions are imported. I don't know. == Hiding Types. == This doesn't even really concern haskell anymore. It's just a question of how you can tell ghci or haddock that it doesn't need to show the contexts that were derived. == Examples == Example1: {{{ a :: Maybe (HasEq a) -> Maybe (HasEq a) -> Bool a (Just x) (Just y) = x == y a _ _ = Nothing }}} according to rule 2. gets translated to. {{{a :: (Eq a) => Maybe (HasEq a) -> Maybe (HasEq a) -> Bool}}} Example2: {{{ b :: Either (HasEq a) (HasShow a) -> String -- what constraints are used here??? b (Left x) = show (x == x) b (Right x) = show x }}} according to rule 2. gets translated to. {{{ b :: (Eq a, Show a) => Either (HasEq a) (HasShow a) -> String }}} Example3: {{{ c :: Proxy (HasEq a) -> () -- what constraints are used here??? c _ = () }}} according to rule 2. gets translated to. Though I don't really know what Proxy does. {{{ c :: (Eq a) => Proxy (HasEq a) -> () }}} {{{ d :: Dynamic -> Maybe Bool d (MkDyn y) | Just (MkHasEq z) <- cast y = Just (z == z) -- where does the Eq constraint come from?? d _ = Nothing }}} according to rule 1. The Eq quality constraint gets set because of the use of the constructor. I'm not sure how to write this in terms of types but it works for GADTs. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8026#comment:19> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #8026: DatatypeContexts should be fixed, not deprecated
by GHC 22 Sep '14

22 Sep '14

#8026: DatatypeContexts should be fixed, not deprecated -------------------------------------+------------------------------------- Reporter: gidyn | Owner: Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.6.3 Component: Compiler | Keywords: Resolution: | Architecture: Unknown/Multiple Operating System: | Difficulty: Unknown Unknown/Multiple | Blocked By: Type of failure: | Related Tickets: None/Unknown | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- Comment (by goldfire): This proposal ''does'' propose a change to the type system, as it proposes a change to the set of well-typed programs. The same is true of partial type signatures. Type inference is an algorithm that implements a way to enforce the constraints of a type system. Your example is clear for that simple situation. But, Haskell's type system is not that simple! For example: {{{ data Eq a => HasEq a = MkHasEq a deriving Typeable data Show a => HasShow a = MkHasShow a a :: Maybe (HasEq a) -> Maybe (HasEq a) -> Bool a (Just x) (Just y) = x == y a _ _ = Nothing b :: Either (HasEq a) (HasShow a) -> String -- what constraints are used here??? b (Left x) = show (x == x) b (Right x) = show x c :: Proxy (HasEq a) -> () -- what constraints are used here??? c _ = () data Dynamic = forall a. Typeable a => MkDyn a d :: Dynamic -> Maybe Bool d (MkDyn y) | Just (MkHasEq z) <- cast y = Just (z == z) -- where does the Eq constraint come from?? d _ = Nothing }}} Your examples seem to address specific, easy cases. But, we need to consider all the wide range of compositions with other Haskell features! -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8026#comment:18> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #9589: Building GHC as a cross-compiler for armhf fails with undefined reference to `dlopen', `dlerror', `dlsym'
by GHC 22 Sep '14

22 Sep '14

#9589: Building GHC as a cross-compiler for armhf fails with undefined reference to `dlopen', `dlerror', `dlsym' -------------------------------------+------------------------------------- Reporter: jpmasseria | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Build System | Version: 7.8.3 Keywords: | Operating System: Linux Architecture: Unknown/Multiple | Type of failure: Building Difficulty: Unknown | GHC failed Blocked By: | Test Case: Related Tickets: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- Running on Ubuntu 14.04.1 I am trying to build a stage1 cross-compiler for a Raspberry Pi running Raspbian. I am using the tool chain located at https://github.com/raspberrypi/tools. I copied my Raspberry Pi's /include and /usr directories to my build machine into directory `/home/johnma/rpi-toolchain/rootfs`. Following these instructions [https://ghcarm.wordpress.com/2014/01/18 /unregisterised-ghc-head-build-for-arm64-platform/] I force gcc to use my sysroot using this bash script: `/home/johnma/mygcc` contains: {{{ #!/bin/bash arm-linux-gnueabihf-gcc --sysroot=/home/johnma/rpi-toolchain/rootfs $@ }}} I used this configure command: {{{ ./configure --target=arm-linux-gnueabihf --enable-unregisterised --with- gcc=/home/johnma/mygcc --with-nm=/usr/bin/nm }}} The build ends with: {{{ "inplace/bin/ghc-stage1" -o utils/hsc2hs/dist-install/build/tmp/hsc2hs -hisuf hi -osuf o -hcsuf hc -static -H64m -O0 -hide-all-packages -i -iutils/hsc2hs/. -iutils/hsc2hs/dist-install/build -iutils/hsc2hs/dist- install/build/autogen -Iutils/hsc2hs/dist-install/build -Iutils/hsc2hs /dist-install/build/autogen -optP-include -optPutils/hsc2hs/dist- install/build/autogen/cabal_macros.h -package base-4.7.0.1 -package containers-0.5.5.1 -package directory-1.2.1.0 -package filepath-1.3.0.2 -package process-1.2.0.0 -XHaskell98 -XCPP -XForeignFunctionInterface -no-user-package-db -rtsopts -odir utils/hsc2hs/dist-install/build -hidir utils/hsc2hs/dist-install/build -stubdir utils/hsc2hs/dist- install/build -optl-L'/home/johnma/ghc/libraries/process/dist- install/build' -optl-L'/home/johnma/ghc/libraries/directory/dist- install/build' -optl-L'/home/johnma/ghc/libraries/unix/dist-install/build' -optl-L'/home/johnma/ghc/libraries/time/dist-install/build' -optl-L'/home/johnma/ghc/libraries/old-locale/dist-install/build' -optl-L'/home/johnma/ghc/libraries/filepath/dist-install/build' -optl-L'/home/johnma/ghc/libraries/containers/dist-install/build' -optl-L'/home/johnma/ghc/libraries/bytestring/dist-install/build' -optl-L'/home/johnma/ghc/libraries/deepseq/dist-install/build' -optl-L'/home/johnma/ghc/libraries/array/dist-install/build' -optl-L'/home/johnma/ghc/libraries/base/dist-install/build' -optl-L'/home/johnma/ghc/libraries/integer-simple/dist-install/build' -optl-L'/home/johnma/ghc/libraries/ghc-prim/dist-install/build' -optl-L'/home/johnma/ghc/rts/dist/build' -optl-lrt -optl-lutil -optl- lpthread -optl-lm -optl-lrt -static -H64m -O0 -hide-all-packages -i -iutils/hsc2hs/. -iutils/hsc2hs/dist-install/build -iutils/hsc2hs/dist- install/build/autogen -Iutils/hsc2hs/dist-install/build -Iutils/hsc2hs /dist-install/build/autogen -optP-include -optPutils/hsc2hs/dist- install/build/autogen/cabal_macros.h -package base-4.7.0.1 -package containers-0.5.5.1 -package directory-1.2.1.0 -package filepath-1.3.0.2 -package process-1.2.0.0 -XHaskell98 -XCPP -XForeignFunctionInterface -no-user-package-db -rtsopts utils/hsc2hs/dist-install/build/Main.o utils/hsc2hs/dist-install/build/HSCParser.o utils/hsc2hs/dist- install/build/DirectCodegen.o utils/hsc2hs/dist- install/build/CrossCodegen.o utils/hsc2hs/dist- install/build/UtilsCodegen.o utils/hsc2hs/dist-install/build/Common.o utils/hsc2hs/dist-install/build/C.o utils/hsc2hs/dist- install/build/Flags.o utils/hsc2hs/dist-install/build/Paths_hsc2hs.o /home/johnma/ghc/rts/dist/build/libHSrts.a(Linker.o): In function `internal_dlopen': Linker.c:(.text+0x68): undefined reference to `dlopen' Linker.c:(.text+0xa4): undefined reference to `dlerror' /home/johnma/ghc/rts/dist/build/libHSrts.a(Linker.o): In function `lookupSymbol': Linker.c:(.text+0x874): undefined reference to `dlerror' Linker.c:(.text+0x89c): undefined reference to `dlsym' Linker.c:(.text+0x8a4): undefined reference to `dlerror' Linker.c:(.text+0x8c4): undefined reference to `dlsym' collect2: error: ld returned 1 exit status make[1]: *** [utils/hsc2hs/dist-install/build/tmp/hsc2hs] Error 1 make[1]: *** Waiting for unfinished jobs.... make: *** [all] Error 2 }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9589> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 9

Re: [GHC] #8026: DatatypeContexts should be fixed, not deprecated
by GHC 22 Sep '14

22 Sep '14

#8026: DatatypeContexts should be fixed, not deprecated -------------------------------------+------------------------------------- Reporter: gidyn | Owner: Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.6.3 Component: Compiler | Keywords: Resolution: | Architecture: Unknown/Multiple Operating System: | Difficulty: Unknown Unknown/Multiple | Blocked By: Type of failure: | Related Tickets: None/Unknown | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- Comment (by gidyn): To emphasise: this proposal is '''not''' proposing a change to the type system. It is, similar to Partial Type Signatures, proposing a new use for GHC's existing type inference capabilities. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8026#comment:17> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #8026: DatatypeContexts should be fixed, not deprecated
by GHC 22 Sep '14

22 Sep '14

#8026: DatatypeContexts should be fixed, not deprecated -------------------------------------+------------------------------------- Reporter: gidyn | Owner: Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.6.3 Component: Compiler | Keywords: Resolution: | Architecture: Unknown/Multiple Operating System: | Difficulty: Unknown Unknown/Multiple | Blocked By: Type of failure: | Related Tickets: None/Unknown | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- Comment (by gidyn): Consider the first example {{{ data Eq a => Foo a = Foo a isEq :: Foo a -> Foo a -> Bool isEq (Foo x) (Foo y) = x == y }}} This won't compile, because `isEq` is missing `Eq a =>`. However, with [https://ghc.haskell.org/trac/ghc/wiki/PartialTypeSignatures#extra- constraints-wildcard Partial Type Signatures], we can write {{{ isEq :: _ => Foo a -> Foo a -> Bool }}} and GHC will infer the constraint. This proposal is requesting that whenever `DatatypeContexts` has been turned on, and a data type appears in a type signature, GHC will automatically infer the data type's context in the same way as it would for an extra-constraints wildcard. Is this clearer? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8026#comment:16> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #5248: Infer type context in a type signature
by GHC 22 Sep '14

22 Sep '14

#5248: Infer type context in a type signature -------------------------------------+------------------------------------- Reporter: gidyn | Owner: Type: feature | Status: closed request | Milestone: 7.10.1 Priority: low | Version: 7.0.3 Component: Compiler | Keywords: (Type checker) | Architecture: Unknown/Multiple Resolution: duplicate | Difficulty: Unknown Operating System: | Blocked By: Unknown/Multiple | Related Tickets: Type of failure: GHC | rejects valid program | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- Changes (by gidyn): * status: new => closed * difficulty: => Unknown * resolution: => duplicate Comment: This is now covered by [https://ghc.haskell.org/trac/ghc/wiki/PartialTypeSignatures#extra- constraints-wildcard Partial Type Signatures] for general type inference, and #8026 for the specific case of data types. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/5248#comment:13> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #8026: DatatypeContexts should be fixed, not deprecated
by GHC 22 Sep '14

22 Sep '14

#8026: DatatypeContexts should be fixed, not deprecated -------------------------------------+------------------------------------- Reporter: gidyn | Owner: Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.6.3 Component: Compiler | Keywords: Resolution: | Architecture: Unknown/Multiple Operating System: | Difficulty: Unknown Unknown/Multiple | Blocked By: Type of failure: | Related Tickets: None/Unknown | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- Comment (by goldfire): It seems that those of us who have hacked on GHC in the past have a hard time understanding this proposal -- which, to me, looks either unsound (if implemented simply) or very complicated (adding the ability to ''return'' a constraint). If you want this to be considered, please write up a wiki page with detailed examples and types. Also, suggest how this feature should be implemented, in terms of a translation to GHC's core language. It currently all seems a little magical. Thanks! -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8026#comment:15> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #8026: DatatypeContexts should be fixed, not deprecated
by GHC 22 Sep '14

22 Sep '14

#8026: DatatypeContexts should be fixed, not deprecated -------------------------------------+------------------------------------- Reporter: gidyn | Owner: Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.6.3 Component: Compiler | Keywords: Resolution: | Architecture: Unknown/Multiple Operating System: | Difficulty: Unknown Unknown/Multiple | Blocked By: Type of failure: | Related Tickets: None/Unknown | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- Changes (by gidyn): * status: closed => new * resolution: wontfix => Comment: See also [https://ghc.haskell.org/trac/ghc/wiki/PartialTypeSignatures #extra-constraints-wildcard Partial Type Signatures]. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8026#comment:14> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0