June 2015 - ghc-tickets

[GHC] #10489: Panic in TcEvidence due to wrong role
by GHC 31 Jul '15

31 Jul '15

#10489: Panic in TcEvidence due to wrong role -------------------------------------+------------------------------------- Reporter: goldfire | Owner: Type: bug | Status: new Priority: highest | Milestone: Component: Compiler | Version: 7.11 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Blocked By: Test Case: | Related Tickets: Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- This single line tickles it: {{{ convert d = let d' = case d of '0' -> '!' in d' }}} produces {{{ ghc: panic! (the 'impossible' happened) (GHC version 7.11.20150604 for x86_64-apple-darwin): ASSERT failed! file compiler/typecheck/TcEvidence.hs line 600 Sym cobox_af8M Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug }}} The failing assertion is about roles. Will look into this tomorrow. Labeling this "highest" because it's a very silly regression. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10489> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 10

[GHC] #9260: Unnecessary error using GHC.TypeLits
by GHC 30 Jul '15

30 Jul '15

#9260: Unnecessary error using GHC.TypeLits -------------------------------------+------------------------------------- Reporter: Iceland_jack | Owner: Type: bug | Status: new Priority: low | Milestone: Component: Compiler | Version: 7.8.2 Keywords: type lits, data | Operating System: Linux kinds, error message | Type of failure: Incorrect Architecture: Unknown/Multiple | warning at compile-time Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | -------------------------------------+------------------------------------- Compiling: {{{ {-# LANGUAGE DataKinds, TypeOperators, GADTs #-} module Error where import GHC.TypeLits data Fin n where Fzero :: Fin (n + 1) Fsucc :: Fin n -> Fin (n + 1) }}} gives a strange (unnecessary) error message claiming that `Fin 1` doesn't match the expected type `Fin (0 + 1)`: {{{ % ghc --version The Glorious Glasgow Haskell Compilation System, version 7.8.2 % ghc -ignore-dot-ghci /tmp/Error.hs [1 of 1] Compiling Error ( /tmp/Error.hs, /tmp/Error.o ) /tmp/Error.hs:12:8: Couldn't match type ‘0’ with ‘1’ Expected type: Fin 1 Actual type: Fin (0 + 1) In the expression: Fsucc Fzero In an equation for ‘test’: test = Fsucc Fzero /tmp/Error.hs:12:14: Couldn't match type ‘1’ with ‘0’ Expected type: Fin 0 Actual type: Fin (0 + 1) In the first argument of ‘Fsucc’, namely ‘Fzero’ In the expression: Fsucc Fzero % }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9260> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #9772: Building documentation alone is broken
by GHC 30 Jul '15

30 Jul '15

#9772: Building documentation alone is broken -------------------------------------+------------------------------------- Reporter: jstolarek | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.10.1 Component: Build System | Version: 7.9 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Building Blocked By: | GHC failed Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- At the moment it seems that the only way to build documentation is to actually build all of GHC, which is nonsense. Methods of building documentation described in Building/Docs are broken: {{{ To build just the Haddock docs for a given library, do this: cd libraries/base make html stage=0 FAST=YES }}} That ends with errors. See `haddock_for_lib.log` attachement. {{{ To build a document on its own, for example the Users Guide, do this: cd docs/users_guide make html stage=0 FAST=YES substitute 'html' for 'pdf' or 'ps' to build other types of documentation. }}} That also ends with errors. See `users_guide.log` attachement. For above experiments I used the following lines in my `mk/build.mk` file: {{{ HADDOCK_DOCS = YES BUILD_DOCBOOK_HTML = YES BUILD_DOCBOOK_PS = NO BUILD_DOCBOOK_PDF = NO }}} Then I run `perl boot && ./configure`. I didn't run `make`, just the above instructions. My configure returns this: {{{ Configure completed successfully. Building GHC version : 7.9.20141105 Build platform : x86_64-unknown-linux Host platform : x86_64-unknown-linux Target platform : x86_64-unknown-linux Bootstrapping using : /home/killy/.ghc-sandbox/bin/ghc which is version : 7.8.3 Using gcc : /usr/bin/gcc which is version : 4.5.1 Building a cross compiler : NO cpp : /usr/bin/gcc cpp-flags : -E -undef -traditional ld : /usr/bin/ld Happy : /home/killy/.cabal/bin/happy (1.19.4) Alex : /home/killy/.cabal/bin/alex (3.1.3) Perl : /usr/bin/perl dblatex : /usr/bin/dblatex xsltproc : /usr/bin/xsltproc Using LLVM tools llc : /usr/bin/llc opt : /usr/bin/opt HsColour : /home/killy/.cabal/bin/HsColour Building DocBook HTML documentation : NO Building DocBook PS documentation : YES Building DocBook PDF documentation : YES }}} I'm setting the milestone to 7.10.1. It's annoying to have this feature of the build system broken and it would be great to have this fixed. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9772> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1

[GHC] #9600: Bad error message: Applicative is not a derivable class
by GHC 30 Jul '15

30 Jul '15

#9600: Bad error message: Applicative is not a derivable class -------------------------------------+------------------------------------- Reporter: nh2 | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.10.1 Component: Compiler | Version: 7.8.3 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Easy (less than 1 | Type of failure: hour) | None/Unknown Blocked By: | Test Case: Related Tickets: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- You have `GeneralizedNewtypeDeriving` on and use `deriving Applicative` with something that you cannot derive Applicative on, GHC correctly says `cannot eta-reduce the representation type enough`. If you don't have `GeneralizedNewtypeDeriving` on yet and deriving Applicative will work, GHC says something like "perhaps you want to enable GeneralizedNewtypeDeriving"? However, if you don't have `GeneralizedNewtypeDeriving` AND if GHC figures out that if you enable it, the next message will be `cannot eta-reduce the representation type enough`, then GHC will state the most unhelpful `Applicative is not a derivable class`, which suggests to me that '''in general''', Applicative is not derivable (which is wrong). rwbarton's suggestion: Suggest enabling `GeneralizedNewtypeDeriving` in any case and then let the user see why it cannot be derived. Some relevant #ghc chat: {{{ rwbarton, that's GND again rwbarton: I a ~ K a where K = ErrorT ImmError (ReaderT CliOptions (ReaderT Config IO)) is an instance of Applicative rwbarton: so it can coerce the instance for K to an instance for I nh2: so is it the error message that is the confusing part here? It suggests that "In general, Applicative is not derivable" nh2: I mean it as: Shouldn't it give an error message like "can't derive an Applicative instance in this case because ..."? rwbarton: well maybe it could add something like "Did you mean to use GeneralizedNewtypeDeriving?" rwbarton: oh I see what you mean, even if you turn that extension on it gives the same error message nh2: yes, because we already have such a message in some cases nh2: no, in this case I had really forgotten the extension. If I add it, I get the "cannot eta-reduce the representation type enough" error, but I haven't figured out why I get it in this case yet rwbarton: I wrote the condition above basically, "I a ~ K a where K = ... is an instance of Applicative" where ~ is the coercion/isomorphism introduced by the newtype rwbarton: if you can't write down K then you can't use GND because where would it get the instance? rwbarton: writing down K is eta reduction nh2: I'm wondering whether it doesn't suggest adding the extension because it already figured out I will not be able to eta-reduce? rwbarton: oh yeah, it seems so rwbarton: that seems a bit silly nh2: in this case, I think we should pimp the error message a bit nh2: (this is the thing I want to derive bwt: https://github.com /ganeti-github-testing/ganeti- test-1/blob/master/src/Ganeti/BasicTypes.hs#L147) nh2: what do you think would be the best error message to show in such a case? nh2: just suggest the GNTD suggestion regardless and let the user see afterwards that it cannot work? rwbarton: I would just have it always suggest GND ... yeah nh2: ok I file a ticket for that rwbarton: seems like more suggestions > fewer suggestions rwbarton: even if this is a weird second-order case where there are two errors }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9600> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #8767: Add rules involving `coerce` to the libraries
by GHC 28 Jul '15

28 Jul '15

#8767: Add rules involving `coerce` to the libraries ------------------------------------+------------------------------------- Reporter: nomeata | Owner: Type: task | Status: new Priority: normal | Milestone: Component: libraries/base | Version: 7.9 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: None/Unknown Difficulty: Unknown | Test Case: Blocked By: 8718 | Blocking: Related Tickets: #2110 | ------------------------------------+------------------------------------- With #2110 fixed, we can now add rules like {{{ {-# RULES "map/coerce" [0] map coerce = coerce #-} }}} to the standard libraries. But probably this should happen together or after #8718. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8767> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 27

[GHC] #8538: confusing specialization CORE warning, also can't mark type class instances INLINEABLE unless class defn is marked INLINEABLE
by GHC 27 Jul '15

27 Jul '15

#8538: confusing specialization CORE warning, also can't mark type class instances INLINEABLE unless class defn is marked INLINEABLE ------------------------------------+------------------------------------- Reporter: carter | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.6.3 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: None/Unknown Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | ------------------------------------+------------------------------------- As part of writing an array library for numerical purposes, I have a static sized strict list type for representing the Indexing tuple. One implementation strategy makes it a tad generic, and then I'm writing a bunch of manipulation utilities using Functor, Applicative and Foldable. I wrote it with all the functions marked INLINABLE so i could ensure that I could have the index arith calculations SPECIALIZE in the common cases. Unfortunately, it seems like type class instances can only marked INLINABLE if the type class definition is itself marked INLINE or INLINABLE. Considering i'm writing a numerical array library, I want to make sure that address arithmetic. This could could be argued to be by design (and thus a wont fix). That aside, I also get the following Warning for my "map2" function, which i found surprising. (and it perhaps is worth some investigation / cleanup ) {{{ src/Numerical/Types/Shape.hs:165:1: Warning: RULE left-hand side too complicated to desugar let { $dFunctor :: Functor (Shape ('S N0)) [LclId] $dFunctor = Numerical.Types.Shape.$fFunctorShape @ 'Z $dFunctor } in map2 @ a @ b @ c @ N2 (Numerical.Types.Shape.$fApplicativeShape @ ('S N0) (Numerical.Types.Shape.$fFunctorShape @ ('S N0) $dFunctor) (Numerical.Types.Shape.$fApplicativeShape @ 'Z $dFunctor Numerical.Types.Shape.$fApplicativeShape0)) }}} {{{ {-# LANGUAGE PolyKinds #-} {-# LANGUAGE BangPatterns #-} {-# LANGUAGE DataKinds #-} {-# LANGUAGE TypeFamilies #-} {-# LANGUAGE GADTs #-} {-# LANGUAGE DeriveDataTypeable #-} {-# LANGUAGE TypeOperators #-} {-# LANGUAGE FlexibleInstances #-} {-# LANGUAGE FlexibleContexts #-} {-# LANGUAGE FunctionalDependencies #-} {-# LANGUAGE UndecidableInstances #-} {-# LANGUAGE ScopedTypeVariables #-} {-# LANGUAGE StandaloneDeriving #-} {-# LANGUAGE CPP #-} {-# LANGUAGE TemplateHaskell #-} {-# LANGUAGE NoImplicitPrelude #-} module Numerical.Types.Shape where import GHC.Magic import Data.Monoid import Data.Functor import Data.Foldable import Control.Applicative -- import Numerical.Types.Nat import Prelude (seq, ($!),($),Show(..),Eq(),Int) data Nat = S !Nat | Z type N0 = Z type N1= S N0 type N2 = S N1 type N3 = S N2 type N4 = S N3 {- not doing the HLIST style shape because I don't want to have any pattern matchings going on. Also would play hell with locality quality in the address translation hackery, because there'd be an extra load to get those ints! -} infixr 3 :* {- the concern basically boils down to "will it specialize / inline well" -} data Shape (rank :: Nat) a where Nil :: Shape Z a (:*) :: !(a) -> !(Shape r a ) -> Shape (S r) a --deriving instance (Show (Shape rank a)) -- deriving instance Eq (Shape rank a) -- #if defined( __GLASGOW_HASKELL__ ) && ( __GLASGOW_HASKELL__ >= 707) --deriving instance Typeable (Shape rank a) -- #endif instance Functor (Shape Z) where fmap = \ f Nil -> Nil {-# INLINABLE fmap #-} {-# SPECIALIZE fmap :: (a ->b )-> (Shape Z a)-> (Shape Z b) #-} instance (Functor (Shape r)) => Functor (Shape (S r)) where fmap = \ f (a :* rest) -> f a :* fmap f rest {-# INLINABLE fmap #-} {-# SPECIALIZE fmap :: (a ->b )-> (Shape N1 a)-> (Shape N1 b) #-} {-# SPECIALIZE fmap :: (a ->b )-> (Shape N2 a)-> (Shape N2 b) #-} instance Applicative (Shape Z) where pure = \ _ -> Nil {-# INLINE pure #-} (<*>) = \ _ _ -> Nil {-# INLINE (<*>) #-} {-# SPECIALIZE pure :: a -> Shape Z a #-} {-# SPECIALIZE (<*>) :: Shape Z (a -> b) -> Shape Z a -> Shape Z b #-} instance Applicative (Shape r)=> Applicative (Shape (S r)) where pure = \ a -> a :* (pure a) {-# INLINE pure #-} (<*>) = \ (f:* fs) (a :* as) -> f a :* (inline (<*>)) fs as {-# INLINE (<*>) #-} {-# SPECIALIZE pure :: a -> (Shape Z a) #-} {-# SPECIALIZE pure :: a -> (Shape N1 a) #-} {-# SPECIALIZE pure :: a -> (Shape N2 a )#-} instance Foldable (Shape Z) where foldMap = \ _ _ -> mempty {-# INLINE foldMap #-} foldl = \ _ init _ -> init {-# INLINE foldl #-} foldr = \ _ init _ -> init {-# INLINE foldr #-} foldr' = \_ !init _ -> init {-# INLINE foldr' #-} foldl' = \_ !init _ -> init {-# INLINE foldl' #-} -- {-# SPECIALIZE foldMap :: Monoid m => (a -> m)-> Shape Z a -> m #-} -- {-# SPECIALIZE foldl :: (a ->b -> a) -> a -> Shape Z b -> a #-} -- {-# SPECIALIZE foldr :: (b ->a -> a) -> a -> Shape Z b -> a #-} -- {-# SPECIALIZE foldl' :: (a ->b -> a) -> a -> Shape Z b -> a #-} -- {-# SPECIALIZE foldr' :: (b ->a -> a) -> a -> Shape Z b -> a #-} instance (Foldable (Shape r)) => Foldable (Shape (S r)) where foldMap = \f (a:* as) -> f a <> foldMap f as {-# INLINE foldMap #-} foldl' = \f !init (a :* as) -> let next = f init a in next `seq` (inline foldl) f next as {-# INLINE foldl' #-} foldr' = \f !init (a :* as ) -> f a $! (inline foldr) f init as {-# INLINABLE foldr' #-} foldl = \f init (a :* as) -> let next = f init a in (inline foldl) f next as {-# INLINE foldl #-} foldr = \f init (a :* as ) -> f a $ (inline foldr) f init as {-# INLINE foldr #-} {-# SPECIALIZE foldMap::Monoid m => (a -> m)-> Shape N1 a -> m #-} {-# SPECIALIZE foldl :: (a ->b -> a) ->a -> Shape N1 b -> a #-} {-# SPECIALIZE foldr :: (b ->a -> a) -> a -> Shape N1 b -> a #-} {-# SPECIALIZE foldl' :: (a ->b -> a) -> a -> Shape N1 b -> a #-} {-# SPECIALIZE foldr' :: (b ->a -> a) -> a -> Shape N1 b -> a #-} {-# SPECIALIZE foldMap :: Monoid m => (a -> m)-> Shape N2 a -> m #-} {-# SPECIALIZE foldl :: (a ->b-> a) ->a -> Shape N2 b -> a #-} {-# SPECIALIZE foldr :: (b ->a -> a) -> a -> Shape N2 b -> a #-} {-# SPECIALIZE foldl' :: (a ->b -> a) -> a -> Shape N2 b -> a #-} {-# SPECIALIZE foldr' :: (b ->a -> a) -> a -> Shape N2 b -> a #-} -- map2 :: (Applicative (Shape r))=> (a->b ->c) -> (Shape r a) -> (Shape r b) -> (Shape r c ) map2 = \f l r -> pure f <*> l <*> r {-# INLINABLE map2 #-} {-# SPECIALIZE map2 :: (a -> b -> c ) -> (Shape Z a) -> (Shape Z b) -> (Shape Z c )#-} {-# SPECIALIZE map2 :: (a -> b -> c ) -> (Shape N1 a) -> (Shape N1 b) -> (Shape N1 c )#-} {-# SPECIALIZE map2 :: (a -> b -> c ) -> (Shape N2 a) -> (Shape N2 b) -> (Shape N2 c )#-} }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8538> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 2

[GHC] #10522: Add UInfixT, like UInfixE or UInfixP but for types
by GHC 27 Jul '15

27 Jul '15

#10522: Add UInfixT, like UInfixE or UInfixP but for types -------------------------------------+------------------------------------- Reporter: spinda | Owner: Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.10.1 Component: Template | Operating System: Unknown/Multiple Haskell | Type of failure: None/Unknown Keywords: | Blocked By: Architecture: | Related Tickets: Unknown/Multiple | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- TemplateHaskell provides [https://hackage.haskell.org/package/template- haskell-2.10.0.0/docs/Language-Haskell-TH-Syntax.html#infix UInfixE and UInfixP] for punting fixity handling of infix operators in expressions and patterns. There doesn't seem to be an equivalent for type operators in TH's [https://hackage.haskell.org/package/template-haskell-2.10.0.0/docs /Language-Haskell-TH-Syntax.html#t:Type type representation], but it would be nice to have one. (In fact, it doesn't appear possible to implement this manually, either, as the TemplateHaskell doesn't expose fixity information at all for type operators, at least as far as I can tell.) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10522> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 6

[GHC] #9982: cross building integer-gmp is running target program on build host
by GHC 26 Jul '15

26 Jul '15

#9982: cross building integer-gmp is running target program on build host -------------------------------------+------------------------------------- Reporter: newsham | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Build | Version: 7.8.4 System | Operating System: Linux Keywords: | Type of failure: Building GHC Architecture: | failed Unknown/Multiple | Blocked By: Test Case: | Related Tickets: Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- I am trying to build ghc-7.8.4 as a cross compiler. After stage1 finishes and it tries to build libraries it detects that it can link against an existing -gmp and gmp.h (which is what I intended) and starts building the libraries/integer-gmp package against it. During the build it builds the inplace/lib/bin/mkGmpDerivedConstants binary for the target (genode) and then runs it on the build host (ubuntu linux), which leads to failure. The error emitted during the build is: "cp" -p libraries/integer- gmp/mkGmpDerivedConstants/dist/build/tmp/mkGmpDerivedConstants inplace/lib/bin/mkGmpDerivedConstants "rm" -f inplace/lib/mkGmpDerivedConstants echo '#!/bin/bash' >> inplace/lib/mkGmpDerivedConstants echo 'executablename="/u1/home/newsham/src/genode/ghc-7.8.4/inplace/lib/bin/mkGmpDerivedConstants"' >> inplace/lib/mkGmpDerivedConstants echo 'datadir="/u1/home/newsham/src/genode/ghc-7.8.4/inplace/lib"' >> inplace/lib/mkGmpDerivedConstants echo 'bindir="/u1/home/newsham/src/genode/ghc-7.8.4/inplace/bin"' >> inplace/lib/mkGmpDerivedConstants echo 'topdir="/u1/home/newsham/src/genode/ghc-7.8.4/inplace/lib"' >> inplace/lib/mkGmpDerivedConstants echo 'pgmgcc="/u1/home/newsham/src/genode/ghc-7.8.4/wrapper-gcc"' >> inplace/lib/mkGmpDerivedConstants echo 'export LD_LIBRARY_PATH=":$LD_LIBRARY_PATH"' >> inplace/lib/mkGmpDerivedConstants echo 'exec "$executablename" ${1+"$@"}' >> inplace/lib/mkGmpDerivedConstants chmod +x inplace/lib/mkGmpDerivedConstants inplace/lib/mkGmpDerivedConstants > libraries/integer- gmp/mkGmpDerivedConstants/dist/GmpDerivedConstants.h inplace/lib/mkGmpDerivedConstants: line 8: /u1/home/newsham/src/genode/ghc-7.8.4/inplace/lib/bin/mkGmpDerivedConstants: No such file or directory make[1]: *** [libraries/integer- gmp/mkGmpDerivedConstants/dist/GmpDerivedConstants.h] Error 127 make[1]: *** Deleting file `libraries/integer- gmp/mkGmpDerivedConstants/dist/GmpDerivedConstants.h' make: *** [all] Error 2 my build environment is large and complex so I'm not providing all details here, but I will be happy to provide any supporting material needed, on request. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9982> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #9267: Lack of type information in GHC error messages when the liberage coverage condition is unsatisfied
by GHC 25 Jul '15

25 Jul '15

#9267: Lack of type information in GHC error messages when the liberage coverage condition is unsatisfied -------------------------+------------------------------------------------- Reporter: | Owner: danilo2 | Status: new Type: bug | Milestone: Priority: | Version: 7.8.2 normal | Operating System: Unknown/Multiple Component: | Type of failure: Incorrect warning at Compiler | compile-time Keywords: | Test Case: Architecture: | Blocking: Unknown/Multiple | Difficulty: | Unknown | Blocked By: | Related Tickets: | -------------------------+------------------------------------------------- Hello! The problem is not a bug related to proper code execution, but it affects the development in such way, we can treat it as a bug. Lets think about such simple code (it can be further simplified of course): {{{#!haskell {-# LANGUAGE FunctionalDependencies #-} {-# LANGUAGE FlexibleInstances #-} {-# LANGUAGE ScopedTypeVariables #-} {-# LANGUAGE UndecidableInstances #-} data UnsafeBase base err val = Value val | Error err | Other (base val) deriving (Show, Eq) class MagicMerge m1 m2 | m1 -> m2 where magicMerge :: m1 a -> m2 a instance MagicMerge (UnsafeBase (UnsafeBase base err) err) (UnsafeBase base err) where magicMerge ma = case ma of Value a -> Value a Error e -> Error e Other o -> o instance MagicMerge (UnsafeBase (UnsafeBase base err1) err2) (UnsafeBase dstBase err1) where magicMerge (ma :: UnsafeBase (UnsafeBase base err1) err2 a) = case ma of Value a -> Value a Error e -> Other $ magicMerge (Error e :: UnsafeBase base err2 a) Other o -> case o of Value a -> Other $ magicMerge (Value a :: UnsafeBase base err2 a) Error e -> Error e Other o' -> Other $ magicMerge (Other o' :: UnsafeBase base err2 a) main = print "help me world!" }}} When trying to compile it using GHC-7.8, we get following error message: {{{#!haskell Illegal instance declaration for ‘MagicMerge (UnsafeBase (UnsafeBase base err1) err2) (UnsafeBase dstBase err1)’ The liberal coverage condition fails in class ‘MagicMerge’ for functional dependency: ‘m1 -> m2’ Reason: lhs type ‘UnsafeBase (UnsafeBase base err1) err2’ does not determine rhs type ‘UnsafeBase dstBase err1’ In the instance declaration for ‘MagicMerge (UnsafeBase (UnsafeBase base err1) err2) (UnsafeBase dstBase err1)’ }}} Which is of course right, but it is completely unhelpfull! When we run the same program against GHC-7.6, we get another message: {{{#!haskell No instance for (MagicMerge (UnsafeBase base err2) dstBase) arising from a use of `magicMerge' Possible fix: add an instance declaration for (MagicMerge (UnsafeBase base err2) dstBase) In the second argument of `($)', namely `magicMerge (Error e :: UnsafeBase base err2 a)' In the expression: Other $ magicMerge (Error e :: UnsafeBase base err2 a) In a case alternative: Error e -> Other $ magicMerge (Error e :: UnsafeBase base err2 a) }}} And it contains a very helpfull information about inferred types. The information is very helpfull, especialy with more complex instances. What is VERY IMPORTANT, when we use the inferred type, the code starts working in both, GHC-7.8 and GHC-7.6. Right now I have to use both versions of GHC to be able to develop my instances faster. So If we add the inferred by GHC-7.6 premise: {{{#!haskell {-# LANGUAGE FunctionalDependencies #-} {-# LANGUAGE FlexibleInstances #-} {-# LANGUAGE ScopedTypeVariables #-} {-# LANGUAGE UndecidableInstances #-} data UnsafeBase base err val = Value val | Error err | Other (base val) deriving (Show, Eq) class MagicMerge m1 m2 | m1 -> m2 where magicMerge :: m1 a -> m2 a instance MagicMerge (UnsafeBase (UnsafeBase base err) err) (UnsafeBase base err) where magicMerge ma = case ma of Value a -> Value a Error e -> Error e Other o -> o instance (MagicMerge (UnsafeBase base err2) dstBase) => MagicMerge (UnsafeBase (UnsafeBase base err1) err2) (UnsafeBase dstBase err1) where magicMerge (ma :: UnsafeBase (UnsafeBase base err1) err2 a) = case ma of Value a -> Value a Error e -> Other $ magicMerge (Error e :: UnsafeBase base err2 a) Other o -> case o of Value a -> Other $ magicMerge (Value a :: UnsafeBase base err2 a) Error e -> Error e Other o' -> Other $ magicMerge (Other o' :: UnsafeBase base err2 a) main = print "help me world!" }}} The code compiles and works ok in both GHC-7.6 and GHC-7.8. In such cases, GHC-7.8 should inform about lacking premises. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9267> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 6

[GHC] #10378: min/max for Double/Float instances are incorrect
by GHC 25 Jul '15

25 Jul '15

#10378: min/max for Double/Float instances are incorrect -------------------------------------+------------------------------------- Reporter: lerkok | Owner: Type: bug | Status: new Priority: high | Milestone: Component: Compiler | Version: 7.10.1 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Blocked By: Test Case: | Related Tickets: Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- This is similar to many other numeric issues around `Double`s and `Float`s. The IEEE754 requires `min` and `max` on floats to return the "other" number, if one of the arguments is `NaN`. The default definitions used in Haskell does not satisfy this property. Furthermore, the current definitions are not commutative, when given `NaNs` and `-0` arguments. The following cases demonstrate the issue with `max`, but `min` has the exact same problems. {{{#!hs Prelude> (0/0) `max` 5 NaN Prelude> 5 `max` (0/0) 5.0 Prelude> isNegativeZero ((-0) `max` 0) False Prelude> isNegativeZero (0 `max` (-0)) True }}} It wouldn't be hard to fix the definitions appropriately; here are reference implementations that are IEEE754 compliant: {{{#!hs maxH x y | isNaN x = y | isNaN y = x | x > y || (x == y && isNegativeZero y) = x | True = y minH x y | isNaN x = y | isNaN y = x | x < y || (x == y && isNegativeZero x) = x | True = y }}} Note that these reference implementations would be quite expensive if GHC were to use them directly. Luckily, IEEE754 compliant min/max operations are supported by all modern CPUs, so doing the "right" thing here is also the cheaper thing as well. (i.e., it'll even be faster than the current incorrect implementation.) On platforms that do not have hardware implementations, the above definitions can serve as the correct implementations, albeit they'll be slower. (Embedded devices, perhaps.) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10378> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 9