July 2015 - ghc-tickets

[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] #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

[GHC] #10681: Teach GHC to interpret all hs files as two types of hs-boot files (abstract types only/full types + values)
by GHC 10 Feb '17

10 Feb '17

#10681: Teach GHC to interpret all hs files as two types of hs-boot files (abstract types only/full types + values) -------------------------------------+------------------------------------- Reporter: ezyang | Owner: ezyang Type: feature | Status: new request | Priority: normal | Milestone: Component: Compiler | Version: 7.11 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Revisions: | -------------------------------------+------------------------------------- This is a new proposal for solving #1409. The big addition here is that we create **two** hs-boot files for each hs file: one that is a full hs-boot file to be imported by hs files to break loops, and a second which only includes abstract types for hs-boot files to import. C.f. #10679 **Discussion.** Here is a slightly goofy `hs-boot` file I've excerpted from GHC: {{{ module Packages where import {-# SOURCE #-} Module (PackageKey) import {-# SOURCE #-} DynFlags (DynFlags) packageKeyPackageIdString :: DynFlags -> PackageKey -> Maybe String }}} The `hs-boot` file must itself import `hs-boot` files, because this boot file is used by `Module` and `DynFlags`; without `{-# SOURCE #-}`, the boot file itself will participate in a cycle! But notice that there is something very interesting: a boot file is ONLY ever interested in importing other modules to get types. Never to import constructors or functions! We can use this observation to give us a mechanical transformation of an `hs` file to an `hs-boot` file, ASSUMING we can define a "second level" of `hs-boot` file to record our abstract types. **Example.** In this example, we have chosen to break the loop from `A`s import to `B`. {{{ module A where import {-# SOURCE #-} B data A = A B f :: A -> Bool f (A (B (A b))) = g b f _ = True module B where import A data B = B A g :: B -> Bool g (B (A (B b))) = f b g _ = False }}} The first-level `hs-boot`s are: {{{ module A where -- not actually used import {-# SOURCE 2 #-} B data A = A B f :: A -> Bool module B where import {-# SOURCE 2 #-} A data B = B A g :: B -> Bool }}} The second-level `hs-boot`s are: {{{ module A where data A module B where -- not actually used data B }}} **Commentary.** Here are some remarks: 1. Because we have to lift the transitive dependencies of anything we `{-# SOURCE #-}` import, it doesn't make sense to have a pragma which explicitly says what to put in the `hs-boot` file; instead, we just put in everything that we *can* handle in an `hs-boot` file (so exclude anything with missing type signatures, type families, etc.) Ideally, these automatic hs-boot files are generated lazily, but they should be reused as necessary. 2. This facility actually makes `{-# SOURCE #-}` a lot more attractive for increasing separate compilation: you can mark an import `{-# SOURCE #-}` to ensure that if its implementation changes, you don't have to recompile this module / you can build the module in parallel with that module. The downside is that when the imported file is modified, we have to regenerate the `hs-boot` stub before we conclude that the types have not changed (as opposed to with separate `hs-boot` files, where a modification to `hs` would not bump the timestamp on `hs-boot`. 3. This seems to definitely suggest that you should never need more than two levels of hs-boot nesting, or perhaps three with kinding. (But maybe someone has a fancy type system feature for which this is not true!) Maybe this applies to signature files too. 4. We can't force the first level of `hs-boot` files to be abstract types, for two reasons: (1) a source file importing the hs-boot file may really need the selector/constructor, and (2) the `hs-boot` files will reflect any cycles from the source files, that's no good! Rolling out to the second level breaks the cycle because abstract types never need any imports. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10681> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 9

[GHC] #8155: Defaulting bug or unfortunate error message with closed type families
by GHC 08 Feb '17

08 Feb '17

#8155: Defaulting bug or unfortunate error message with closed type families ------------------------------------+------------------------------------- Reporter: nh2 | 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: | ------------------------------------+------------------------------------- https://github.com/nh2/infinite-type-families/blob/master/Test2.hs Some code extracted from hmatrix: {{{ {-# LANGUAGE CPP #-} {-# LANGUAGE TypeFamilies #-} {-# LANGUAGE FlexibleInstances #-} module Test where data Vector a = Vector a data Matrix a = Matrix a class Build f where build' :: BoundsOf f -> f -> ContainerOf f #if __GLASGOW_HASKELL__ <= 706 -- Normal type families with GHC 7.6, works well. type family BoundsOf x type instance BoundsOf (a->a) = Int type instance BoundsOf (a->a->a) = (Int,Int) type family ContainerOf x type instance ContainerOf (a->a) = Vector a type instance ContainerOf (a->a->a) = Matrix a #else -- Closed type families for GHC 7.8 type family BoundsOf x where BoundsOf (a->a->a) = (Int,Int) BoundsOf (a->a) = Int type family ContainerOf x where ContainerOf (a->a) = Vector a ContainerOf (a->a->a) = Matrix a #endif instance (Num a) => Build (a->a) where build' = buildV buildV :: (Integral a, Num b) => a -> (b -> c) -> Vector c buildV _ _ = undefined }}} This is probably related to #8154. With GHC 7.7, we get the following error: {{{ Test2.hs:59:14: Could not deduce (BoundsOf (a -> a) ~ Integer) from the context (Num a) bound by the instance declaration at Test2.hs:58:10-32 Expected type: BoundsOf (a -> a) -> (a -> a) -> ContainerOf (a -> a) Actual type: Integer -> (a -> a) -> Vector a Relevant bindings include build' :: BoundsOf (a -> a) -> (a -> a) -> ContainerOf (a -> a) (bound at Test2.hs:59:5) In the expression: buildV In an equation for ‛build'’: build' = buildV }}} 1) Where does this Integer come from? 2) Change {{{ BoundsOf (a->a) = Int }}} into {{{ BoundsOf (a->a) = Integer }}} You get the same error message, although clearly {{{BoundsOf (a -> a) ~ Integer}}} (you just wrote that down).[[BR]] Even if the order matters in closed type families, is this not the wrong error message? 3) If you flip the order so that the (a->a) case comes first, it works (with both it Int or Integer). I don't quite understand why. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8155> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 8

[GHC] #10635: -fwarn-redundant-constraints should not be part of -Wall
by GHC 07 Feb '17

07 Feb '17

#10635: -fwarn-redundant-constraints should not be part of -Wall -------------------------------------+------------------------------------- Reporter: Lemming | Owner: Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.11 Component: Compiler | Operating System: Unknown/Multiple (Type checker) | Type of failure: Incorrect Keywords: | warning at compile-time Architecture: | Blocked By: Unknown/Multiple | Related Tickets: Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- When I compile existing code with GHC-7.11.20150707 I get lot of "redundant constraints" warnings. Generally I think that this warning can be very useful for minimizing constraints, nonetheless there are good reasons not choose minimal constraints for a type signature. E.g. if I have to implement `Data.Map.singleton` I do not need the `Ord k` constraint for the key type. However, I might add it anyway in order to be able to change the implementation in future such that it uses `Ord k` dictionary. Thus I suggest to exclude `-fwarn-redundant-constraints` from `-Wall`. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10635> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 38

[GHC] #10052: Panic (something to do with floatExpr?)
by GHC 31 Jan '17

31 Jan '17

#10052: Panic (something to do with floatExpr?) -------------------------------------+------------------------------------- Reporter: edsko | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.1-rc2 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Blocked By: Test Case: | Related Tickets: Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- Loading {{{ main = let (x :: String) = "hello" in putStrLn x }}} using a very simple driver for the GHC API (see T145.hs) causes a ghc panic: {{{ [1 of 1] Compiling Main ( T145-input.hs, interpreted ) T145: T145: panic! (the 'impossible' happened) (GHC version 7.10.0.20150128 for x86_64-apple-darwin): floatExpr tick <<details unavailable>> Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug }}} This panic is arising in our test case for #8333, so it may be related to that bug. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10052> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 20

[GHC] #10420: "Care with plugin imports" is wrong / orphan RULE visibility
by GHC 29 Jan '17

29 Jan '17

#10420: "Care with plugin imports" is wrong / orphan RULE visibility -------------------------------------+------------------------------------- Reporter: ezyang | Owner: ezyang Type: bug | Status: new Priority: low | Milestone: Component: Compiler | Version: 7.11 (Type checker) | Operating System: Unknown/Multiple Keywords: | Type of failure: None/Unknown Architecture: | Blocked By: Unknown/Multiple | Related Tickets: Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- A module loaded by a plugin poisons the module cache, so we never load an orphan RULE or instance even if we legitimately should do so. The way to test this is a bit convoluted, but it goes something like this: plugins07.hs {{{ module Main where import Plugins07a import RuleDefiningPlugin {-# NOINLINE x #-} x = "foo" main = putStrLn (show x) }}} Plugins07a.hs {{{ --{-# OPTIONS_GHC -fplugin RuleDefiningPlugin #-} module Plugins07a where }}} RuleDefiningPlugin.hs, in ANOTHER PACKAGE (otherwise, EPT rules don't apply) {{{ module RuleDefiningPlugin where import GhcPlugins {-# RULES "unsound" forall x. show x = "SHOWED" #-} plugin :: Plugin plugin = defaultPlugin }}} I'll commit the full test. Here's what happens: * Building `Plugins07a` results in `loadPluginInterface` on `RuleDefiningPlugin`. We load the `ModIface` but only add its types to the environment because of the "Care with plugin imports" special case. * Building `plugins07.hs` results in a normal source level import for `RuleDefiningPlugin`, but `ModIface` is already in the cache so we don't load anything. RULE is not loaded, disaster! Admittedly, actually triggering this bug requires a convoluted chain of events. But really this problem arose because the "Care with plugin imports" fix is just completely nonsense. Here's what we should do: * We should apply the same fix from #2182 on orphan instances to orphan rules too. This way, we can safely load RULEs into the EPS without accidentally bringing them into scope when they shouldn't be. * Loading an interface should unconditionally suck in the instances and rules. The result is more correct and simpler, so it seems worth fixing. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10420> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 8

[GHC] #9351: add ability to version symbols .c for packages with C code
by GHC 27 Jan '17

27 Jan '17

#9351: add ability to version symbols .c for packages with C code -------------------------------------+------------------------------------- Reporter: slyfox | Owner: Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.8.3 Keywords: backpack | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Blocked By: | None/Unknown Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- Let's consider an example: We have a haskell package with some C code imported in haskell: {{{ // a.h file int get_something(void); // a.c file #include "a.h" int get_something(void) { return 42; } // M.hs file module M where foreign import ccall "a.h get_something" :: IO Int }}} The problem here is we can't mix that package with other packages having global 'get_something' symbol. Sometimes it happens when a package copies part of implementation from another package with .c bits under different haskell namespace. Haskell parts would coexist freely, but not C symbols. Would be great if ghc would export some unique package identifier in a way C code could attach it to all exported symbols making linking possible. Something like that: {{{ // a.h file #include "HsFFI.h" /* #define FOR_HASKELL(__sym) package_id_##__sym */ int FOR_HASKELL(get_something)(void); // a.c file #include "a.h" int FOR_HASKELL(get_something)(void) { return 42; } // M.hs file module M where foreign import ccall-for-haskell "a.h get_something" :: IO Int }}} That way we explicitly mark symbol as inaccessible to external plain C code. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9351> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #8901: (very) bad inline heuristics
by GHC 25 Jan '17

25 Jan '17

#8901: (very) bad inline heuristics ------------------------------------+------------------------------------- Reporter: heisenbug | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.7 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: None/Unknown Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | ------------------------------------+------------------------------------- I observe a bad case for inlining blowup for the time library. Observed for 7.7, just checked in 7.9. The problem is so bad that the PowerPC32 assembler cannot resolve cross- routine references. (I have no error message at hand, but could try to get one.) Instead here are the stats for x86, with inlining and certain inlining suppressed. The architecture seems irrelevant. {{{ $ file ./libraries/time/dist-install/build/Data/Time/Format.o ./libraries/time/dist-install/build/Data/Time/Format.o: Mach-O object i386 $ ls -l ./libraries/time/dist-install/build/Data/Time/Format.* -rw-r--r-- 1 ggreif staff 61875 Mar 15 16:48 ./libraries/time/dist- install/build/Data/Time/Format.dyn_hi -rw-r--r-- 1 ggreif staff 450864 Mar 15 16:48 ./libraries/time/dist- install/build/Data/Time/Format.dyn_o -rw-r--r-- 1 ggreif staff 61863 Mar 15 16:48 ./libraries/time/dist- install/build/Data/Time/Format.hi -rw-r--r-- 1 ggreif staff 332216 Mar 15 16:48 ./libraries/time/dist- install/build/Data/Time/Format.o }}} With following patch applied: <https://github.com/ggreif/packages- time/commit/83f08b8896b950707a910fed4e30bddb7ee7f52f> I get: {{{ $ ls -l ./libraries/time/dist-install/build/Data/Time/Format.* -rw-r--r-- 1 ggreif staff 10554 Mar 16 00:08 ./libraries/time/dist- install/build/Data/Time/Format.dyn_hi -rw-r--r-- 1 ggreif staff 106832 Mar 16 00:08 ./libraries/time/dist- install/build/Data/Time/Format.dyn_o -rw-r--r-- 1 ggreif staff 10542 Mar 16 00:08 ./libraries/time/dist- install/build/Data/Time/Format.hi -rw-r--r-- 1 ggreif staff 78864 Mar 16 00:08 ./libraries/time/dist- install/build/Data/Time/Format.o }}} The factor is 4.2 for Format.o, Format.dyn_o and 5.8 for the .hi files. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8901> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 12

[GHC] #8156: amd64 + in-tree gmp build broken
by GHC 23 Jan '17

23 Jan '17

#8156: amd64 + in-tree gmp build broken -------------------------------------+------------------------------------- Reporter: errge | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.8.1 Component: Build System | Version: 7.7 Keywords: | Operating System: Unknown/Multiple Architecture: x86_64 (amd64) | Type of failure: Building GHC Difficulty: Easy (less than 1 | failed hour) | Test Case: Blocked By: | Blocking: Related Tickets: #4022 #4366 #4374 | -------------------------------------+------------------------------------- Currently, the GHC build system has an in-tree GMP. This GMP seems to be used only if there is no usable GMP found on the system and the user didn't provide one using configure flags. When we build GHC using an external shared version of GMP, the resulting GHC is dynamically linked against GMP and also the generated programs are linked dynamically against GMP. When the in-tree GMP is used, it is built statically, even if the GHC we are building is shared GHC on GNU/Linux. This feature is currently broken on amd64 (#4366, #4022), but is working currently very well on Linux i686. Apart from the breakage this setting is very flexible and I'd be very happy if it were possible to keep it, in spite of removeal requests like #4374. When there is a system GMP, nothing would change, this is the most typical usage, e.g. distributions are all building this way. On the other hand, if the user removes his GMP (maybe we should provide an option to explicity trigger this in ./configure, instead of having to remove GMP); she can build a GHC that is not linked dynamically against GMP and the resulting programs are not linked either. This makes it possible to generate binaries that link dynamically against libc (no -static) which is quite compatible amongst versions, but doesn't require a libgmp.so to run. For binary deployments this seems to be the best kind of setup nowadays. Now, if we were only able to fix this for amd64. See this: https://www.gentoo.org/proj/en/base/amd64/howtos/index.xml?part=1&chap=3 On amd64 we can continue to compile GMP statically and link into the binary library, we just have to compile GMP with the fPIC. I attached the patch to fix this. Previously a similar patch has been rejected in #4022 because "this shouldn't be platform specific", but this is clearly not the case, this is specific to amd64. Tadaaaam: {{{ errge@sid64:~/ghc/inplace/bin$ cat test.hs main = print (2^100) errge@sid64:~/ghc/inplace/bin$ ./ghc-stage2 test.hs [1 of 1] Compiling Main ( test.hs, test.o ) Linking test ... errge@sid64:~/ghc/inplace/bin$ ./test 1267650600228229401496703205376 errge@sid64:~/ghc/inplace/bin$ ldd ./test linux-vdso.so.1 (0x00007fff79962000) libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007fe4b9a83000) librt.so.1 => /lib/x86_64-linux-gnu/librt.so.1 (0x00007fe4b987a000) libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007fe4b9676000) libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007fe4b92c6000) libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 (0x00007fe4b90a6000) /lib64/ld-linux-x86-64.so.2 (0x00007fe4b9d83000) }}} No GMP! I attach the build file patch to the bugreport. So, can we please apply this patch and just keep the in-tree GMP? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8156> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 11