August 2016 - ghc-tickets

[GHC] #9432: IncoherentInstances are too restricted
by GHC 15 Feb '17

15 Feb '17

#9432: IncoherentInstances are too restricted -------------------------------------+------------------------------------- Reporter: danilo2 | Owner: Type: feature request | Status: new Priority: high | Milestone: Component: Compiler (Type | Version: 7.8.2 checker) | Operating System: Keywords: | Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: Difficulty: Unknown | None/Unknown Blocked By: | Test Case: Related Tickets: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- Hello! The reason behind using `IncoherentInstances`is that we want GHC to commit to a specific instance even if after instantiation of some variables, other instance could be more specific (http://www.haskell.org/ghc/docs/7.0.4/html/users_guide/type-class- extensions.html). There is one problem though. Lets consider following example (which could not make much sense, but shows the problem well): {{{#!haskell {-# OPTIONS_GHC -fno-warn-missing-methods #-} {-# LANGUAGE NoMonomorphismRestriction #-} {-# LANGUAGE FlexibleInstances #-} {-# LANGUAGE PolyKinds #-} {-# LANGUAGE DeriveDataTypeable #-} {-# LANGUAGE TypeFamilies #-} {-# LANGUAGE IncoherentInstances #-} {-# LANGUAGE ScopedTypeVariables #-} module Main where import Data.Typeable -------------------------------------------------------------------------------- -- Type classes -------------------------------------------------------------------------------- class InferType a where inferType :: Proxy a -> Proxy a inferType = undefined -------------------------------------------------------------------------------- -- Proxy datatypes -------------------------------------------------------------------------------- data Id0 = Id0 deriving (Show, Typeable) data Id1 t1 = Id1 deriving (Show, Typeable) data Id2 t1 t2 = Id2 deriving (Show, Typeable) data Id3 t1 t2 t3 = Id3 deriving (Show, Typeable) data Id4 t1 t2 t3 t4 = Id4 deriving (Show, Typeable) data Id5 t1 t2 t3 t4 t5 = Id5 deriving (Show, Typeable) -------------------------------------------------------------------------------- -- Instances -------------------------------------------------------------------------------- instance InferType Int instance (InferType m, InferType a) => InferType (m a) instance (a~Id0) => InferType (a :: *) instance (a~Id1) => InferType (a :: * -> *) instance (a~Id2) => InferType (a :: * -> * -> *) -------------------------------------------------------------------------------- -- Tests -------------------------------------------------------------------------------- toProxy :: a -> Proxy a toProxy _ = Proxy --inferTypeBase :: a -> Proxy a inferTypeBase a = inferType $ toProxy a instance InferType Foo1 instance InferType Foo2 tm _ = 5 data Foo1 a = Foo1 a deriving (Show, Typeable) data Foo2 a b = Foo2 a b deriving (Show, Typeable) instance Monad Id1 -- dummy instances just for tests instance Num Id0 main = do print $ typeOf $ inferType $ toProxy $ (return (5)) print $ typeOf $ inferType $ toProxy $ (5) print $ typeOf $ inferType $ toProxy $ (Foo1 (return 5)) print $ typeOf $ inferType $ toProxy $ (Foo2 (return 5) (return 5)) print "hello" ---- OUTPUT: -- Proxy (Id1 Id0) -- Proxy Id0 -- Proxy (Foo1 (Id1 Id0)) -- Proxy (Foo2 (Id1 Id0) (Id1 Id0)) -- "hello" }}} The idea is very simple - replace all unknown type variables with known ones. It works great, but the problem appears with the function `inferTypeBase`. It should have type of `a -> Proxy a`, but GHC claims it has got `Id0 -> Proxy Id0`. It is of course caused by enabled `-XIncoherentInstances` flag, but I think GHC should be more liberal here. If it really cannot pick any instance (causing an error using only `OverlappingInstances`), the `IncoherentInstances` should allow it to pick matching one even if more specific could be available after instantianization. But If no error would occur while using `OverlappingInstances`, `IncoherentInstances` should not affect the way it resolves the instances. I think this would make `IncoherentInstances` much more useful than now. Maybe it would be good to just add some options to the flag? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9432> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 8

[GHC] #11031: Record Pattern Synonym Cleanup
by GHC 14 Feb '17

14 Feb '17

#11031: Record Pattern Synonym Cleanup -------------------------------------+------------------------------------- Reporter: mpickering | Owner: Type: task | Status: new Priority: low | Milestone: Component: Compiler | Version: 7.10.2 Keywords: newcomer | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Ben comments that there are some loose ends with record pattern synonyms. * It's still not clear to me that the free variables produced in rnPatSynBind are correct * Perhaps in the future we want to refactor tc_single to use a function which extends the typechecking environment instead of using setGblEnv * @mpickering's suggested refactoring of RecSelId to RecSelId (Either PatSyn TyCon) Bool described in the comment in TcExpr.hs * There is a TODO in tc_patsyn_finish which should either be clarified so it can be considered actionable, removed, or just fixed The comment from TcExpr was: > I tried to refactor RecSelId to RecSelId [DataCon] Bool but this causes some pain when it comes to the interface files. Next try it is something like RecSelId (Either PatSyn TyCon) Bool with the aim for something better in the future. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11031> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #9562: Type families + hs-boot files = unsafeCoerce
by GHC 13 Feb '17

13 Feb '17

#9562: Type families + hs-boot files = unsafeCoerce -------------------------------------+------------------------------------- Reporter: goldfire | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.8.3 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: GHC Blocked By: | accepts invalid program Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- Consider the following bundle of modules: A.hs: {{{#!hs {-# LANGUAGE TypeFamilies #-} module A where type family F a b }}} B.hs-boot: {{{#!hs module B where import A oops :: F a b -> a -> b }}} B.hs: {{{#!hs {-# LANGUAGE TypeFamilies #-} module B where import A import C type instance F a b = b oops :: F a b -> a -> b oops = const }}} C.hs: {{{#!hs module C (oops) where import {-# SOURCE #-} B }}} D.hs: {{{#!hs {-# LANGUAGE TypeFamilies #-} module D where import A import C type instance F a b = a unsafeCoerce :: a -> b unsafeCoerce x = oops x x }}} Main.hs: {{{#!hs module Main where import D ( unsafeCoerce ) main = print $ (unsafeCoerce True :: Int) }}} When loading these into GHCi, we quite reasonably get a type family instance overlap error. But, separate compilation leads to disaster: {{{ rae:01:49:47 ~/temp/bug> ghc --version The Glorious Glasgow Haskell Compilation System, version 7.8.3 rae:01:49:49 ~/temp/bug> ghc -c A.hs rae:01:49:53 ~/temp/bug> ghc -c B.hs-boot rae:01:49:58 ~/temp/bug> ghc -c C.hs rae:01:50:09 ~/temp/bug> ghc -c B.hs rae:01:50:13 ~/temp/bug> ghc -c D.hs rae:01:50:17 ~/temp/bug> ghc Main.hs -o Unsafe [6 of 6] Compiling Main ( Main.hs, Main.o ) Linking Unsafe ... rae:01:50:23 ~/temp/bug> ./Unsafe 2882303761534249061 }}} Yikes! Proposed (terrible) solution: hs-boot files '''must''' list all type instance declarations in the corresponding modules. It may also be a good idea to require all normal instance declarations in the hs-boot file as well, because this same trick can be used to introduce incoherence (I think -- haven't tested). This bug persists even if `Main` declares that it is `Safe`. I've attached a tarball of the files for ease of testing. (Credit to Edward Yang and Geoff Mainland, whose discussion provoked the line of inquiry that led to this discovery.) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9562> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 15

[GHC] #8405: experiment with using function-sections for linking (for smaller libs and executables)
by GHC 11 Feb '17

11 Feb '17

#8405: experiment with using function-sections for linking (for smaller libs and executables) ------------------------------------+------------------------------------- Reporter: carter | Owner: Type: task | Status: new Priority: normal | Milestone: 7.10.1 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: | ------------------------------------+------------------------------------- currently the only support we have for making small (ish) executables using split objects, which 1. is not on by default, for various good reasons 2. can seriously penalize compile times 3. in tandem, something like strip is often used, but on some platforms/targets (such as the iOS cross compiler), strip will actually break things might be worth seriously experimenting with function-sections / gc- sections in the object files we generate. example slide deck on this http://elinux.org/images/2/2d/ELC2010-gc- sections_Denys_Vlasenko.pdf -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8405> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 21

[GHC] #12423: Panic with DeriveAnyClass and DefaultSignatures
by GHC 10 Feb '17

10 Feb '17

#12423: Panic with DeriveAnyClass and DefaultSignatures -------------------------------------+------------------------------------- Reporter: knrafto | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- With the code {{{#!hs {-# LANGUAGE DefaultSignatures, DeriveAnyClass #-} class Eq1 f where (==#) :: Eq a => f a -> f a -> Bool default (==#) :: Eq (f a) => f a -> f a -> Bool (==#) = (==) data Foo a = Foo (Either a a) deriving (Eq, Eq1) }}} GHC 8.0.1 and 7.10.3 both panic: {{{ ghc: panic! (the 'impossible' happened) (GHC version 8.0.1 for x86_64-apple-darwin): in other argument Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12423> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 9

[GHC] #9821: DeriveAnyClass support for higher-kinded classes + some more comments
by GHC 10 Feb '17

10 Feb '17

#9821: DeriveAnyClass support for higher-kinded classes + some more comments -------------------------------------+------------------------------------- Reporter: dreixel | Owner: dreixel Type: bug | Status: new Priority: normal | Milestone: 7.10.1 Component: Compiler | Version: 7.9 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Blocked By: | None/Unknown Related Tickets: #5462 | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- #5462 was merged before a few final fixes were done. This ticket serves to record that. Left to address are: https://phabricator.haskell.org/D476?id=1443#inline-3652 https://phabricator.haskell.org/D476?id=1443#inline-3648 https://phabricator.haskell.org/D476?id=1443#inline-3646 Also, `DeriveAnyClass` currently panics at higher-kinded classes (like `((* -> *) -> *) -> Constraint`). I'll fix this. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9821> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 13

[GHC] #12144: GHC panic when using DeriveAnyClass with functor-like class and datatype with a type variable in a contravariant position
by GHC 10 Feb '17

10 Feb '17

#12144: GHC panic when using DeriveAnyClass with functor-like class and datatype with a type variable in a contravariant position -------------------------------------+------------------------------------- Reporter: RyanGlScott | Owner: Type: bug | Status: new Priority: normal | Milestone: 8.2.1 Component: Compiler | Version: 8.0.1 Keywords: Generics | Operating System: Unknown/Multiple Architecture: | Type of failure: Compile-time Unknown/Multiple | crash Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- The following code {{{#!hs {-# LANGUAGE DeriveAnyClass #-} {-# LANGUAGE KindSignatures #-} module Example where class C (a :: * -> *) data T a = MkT (a -> Int) deriving C }}} fails to compile with the following GHC panic: {{{ $ /opt/ghc/8.0.1/bin/ghc Example.hs [1 of 1] Compiling Example ( Example.hs, Example.o ) ghc: panic! (the 'impossible' happened) (GHC version 8.0.1 for x86_64-unknown-linux): contravariant Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug }}} The culprit is that when `DeriveAnyClass` comes up with the constraints for typeclass instances whose argument is of kind `* -> *`, it calls [http://git.haskell.org/ghc.git/blob/0676e68cf5fe8696f1f760fef0f35dba14db110… deepSubtypesContaining], which immediately bails out when it sees a type variable in a contravariant or "bad" position. I'm less sure of what the fix should be. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12144> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #12503: Template Haskell regression: GHC erroneously thinks a type variable is also a kind
by GHC 10 Feb '17

10 Feb '17

#12503: Template Haskell regression: GHC erroneously thinks a type variable is also a kind -------------------------------------+------------------------------------- Reporter: RyanGlScott | Owner: Type: bug | Status: new Priority: high | Milestone: Component: Template | Version: 8.0.1 Haskell | Keywords: TypeInType | Operating System: Unknown/Multiple Architecture: | Type of failure: GHC rejects Unknown/Multiple | valid program Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- The following program compiles without issue on GHC 7.6.3 through GHC 7.10.3, but fails to compile on GHC 8.0.1 and GHC HEAD. (I added a CPP guard simply because the `DataD` constructor changed between 7.10 and 8.0.) {{{#!hs {-# LANGUAGE CPP #-} {-# LANGUAGE KindSignatures #-} {-# LANGUAGE PolyKinds #-} {-# LANGUAGE TemplateHaskell #-} {-# OPTIONS_GHC -ddump-splices #-} module Regression where import Language.Haskell.TH data T k class C a $(do TyConI (DataD [] tName [ KindedTV kName kKind] #if __GLASGOW_HASKELL__ >= 800 _ #endif _ _) <- reify ''T d <- instanceD (cxt []) (conT ''C `appT` (conT tName `appT` sigT (varT kName) kKind)) [] return [d]) }}} {{{ $ /opt/ghc/8.0.1/bin/ghc Regression.hs [1 of 1] Compiling Regression ( Regression.hs, Regression.o ) Regression.hs:(13,3)-(19,15): Splicing declarations do { TyConI (DataD [] tName_a2RT [KindedTV kName_a2RU kKind_a2RV] _ _ _) <- reify ''T; d_a31Y <- instanceD (cxt []) (conT ''C `appT` (conT tName_a2RT `appT` sigT (varT kName_a2RU) kKind_a2RV)) []; return [d_a31Y] } ======> instance C (T (k_avB :: k_avC)) Regression.hs:13:3: error: Variable ‘k_avB’ used as both a kind and a type Did you intend to use TypeInType? }}} Somewhat confusingly, you can use `instance C (T (k_avB :: k_avC))` on its own, and it will compile without issue. Also, quoting it doesn't seem to trip up this bug, as this also compiles without issue: {{{#!hs {-# LANGUAGE CPP #-} {-# LANGUAGE KindSignatures #-} {-# LANGUAGE PolyKinds #-} {-# LANGUAGE TemplateHaskell #-} {-# OPTIONS_GHC -ddump-splices #-} module WorksSomehow where import Language.Haskell.TH data T k class C a $([d| instance C (T k) |]) }}} The original program has several workarounds: 1. Turn off `PolyKinds` (OK, this isn't a very good workaround...) 2. Add an explicit kind signature to `T`, e.g., `data T (k :: k1)` 3. Change the type variable of `T` to some other letter, e.g., `data T p` or `data T k1` Given that (3) is a successful workaround, I strongly suspect that GHC is confusing the type variable `k` (which gets `ddump-splice`d as `k_avB`) with the kind variable `k` (which gets `ddump-splice`d as `k_avC`) due to their common prefix `k`. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12503> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 12

[GHC] #11668: SPEC has a runtime cost
by GHC 10 Feb '17

10 Feb '17

#11668: SPEC has a runtime cost -------------------------------------+------------------------------------- Reporter: bgamari | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.3 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: Runtime Unknown/Multiple | performance bug Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- As far as I know, `SPEC` is merely intended to help the compiler identify calls for which constructor specialization may be help. Yet it's not erased! In fact, it has a runtime representation which we need to construct and pass around to so-marked calls. This likely isn't killing the performance of anyone's programs but surely we can do better than this. Really, the whole scheme of marking functions with `SPEC` (which as far as I know isn't even properly documented) seems quite odd. What is stopping us from simply attaching pragmas to equations? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11668> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #10626: Missed opportunity for SpecConstr
by GHC 10 Feb '17

10 Feb '17

#10626: Missed opportunity for SpecConstr -------------------------------------+------------------------------------- Reporter: simonpj | Owner: Type: bug | Status: new Priority: normal | 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: | -------------------------------------+------------------------------------- Look at `perf/should_run/T4830`. After `SpecConstr` and optimisation we finally get {{{ Rec { Main.foo_$s$wfoo1 [Occ=LoopBreaker] :: Int# -> Double -> Double -> Double# [GblId, Arity=3, Caf=NoCafRefs, Str=DmdType <L,1*U><L,U><L,U>] Main.foo_$s$wfoo1 = \ (sc_s4eN :: Int#) (sc1_s4eO :: Double) (sc2_s4eP :: Double) -> case sc_s4eN of ds_X1SR { __DEFAULT -> case sc1_s4eO of wild_a1Tf { D# x_a1Th -> case sc2_s4eP of wild1_a1Tj { D# y_a1Tl -> case tagToEnum# @ Bool (<=## x_a1Th y_a1Tl) of _ [Occ=Dead] { False -> Main.foo_$s$wfoo1 (-# ds_X1SR 1#) wild1_a1Tj wild_a1Tf; True -> Main.foo_$s$wfoo1 (-# ds_X1SR 1#) wild_a1Tf wild1_a1Tj } } }; 0# -> case sc1_s4eO of _ [Occ=Dead] { D# x_a1UL -> case sc2_s4eP of _ [Occ=Dead] { D# y_a1UP -> +## x_a1UL y_a1UP } } } end Rec } }}} If we ran `SpecConstr` again we'd specialise this function, because the recursive calls both have boxed arguments. I looked into why `SpecConstr` didn't catch it, and it's because `SpecConstr`'s input looks like this {{{ case case tagToEnum# @ Bool (<=## x_a1Th y_a1Tl) of _ [Occ=Dead] { False -> (wild1_a1Tj, wild_a1Tf); True -> wild_X8 } of r_s4e0 { (ipv_s4e1, ipv_s4e2) -> $wfoo_s4db (I# (-# ds_X1SR 1#)) (Just @ (Double, Double) r_s4e0) }}} Notice the case-of-case which doesn't expose the `Double` boxes of arguments to `$wfoo`. Why is that case-of-case still there? Because of `Note [Single- alternative cases]` in Simplify. Which is clearly a delicate spot so I don't want to meddle with it today. But it's intriguing. Maybe Sequent Core will do better. Note that late demand analysis also catches this case, yielding the (rather good) {{{ Rec { Main.$w$s$wfoo [InlPrag=[0], Occ=LoopBreaker] :: Int# -> Double# -> Double# -> Double# [GblId, Arity=3, Caf=NoCafRefs, Str=DmdType <S,1*U><L,U><L,U>] Main.$w$s$wfoo = \ (w_s4gW :: Int#) (ww_s4h1 :: Double#) (ww1_s4h5 :: Double#) -> case w_s4gW of ds_X1SW { __DEFAULT -> case tagToEnum# @ Bool (<=## ww_s4h1 ww1_s4h5) of _ [Occ=Dead] { False -> Main.$w$s$wfoo (-# ds_X1SW 1#) ww1_s4h5 ww_s4h1; True -> Main.$w$s$wfoo (-# ds_X1SW 1#) ww_s4h1 ww1_s4h5 }; 0# -> +## ww_s4h1 ww1_s4h5 } end Rec } }}} I'm pretty sure that late-demand-analysis should be on with `-O2` but someone should do a nofib run to check. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10626> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 10