June 2015 - ghc-tickets

Re: [GHC] #7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)?
by GHC 23 Jun '15

23 Jun '15

#7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)? -------------------------------------+------------------------------------- Reporter: iustin | Owner: bgamari Type: bug | Status: new Priority: normal | Milestone: 7.12.1 Component: Compiler | Version: 7.6.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple performance bug | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Revisions: Phab:D1012 -------------------------------------+------------------------------------- Changes (by bgamari): * owner: => bgamari * differential: => Phab:D1012 Comment: Seems to work. Opened a Differential. We'll see how it looks. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7450#comment:17> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)?
by GHC 23 Jun '15

23 Jun '15

#7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)? -------------------------------------+------------------------------------- Reporter: iustin | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.12.1 Component: Compiler | Version: 7.6.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple performance bug | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Revisions: -------------------------------------+------------------------------------- Comment (by bgamari): Indeed most of the time spent in the specialiser is spent in `consDictBind`. With respect to `snocDictBinds`: I believe `final_binds` in `specBind` actually can be of length greater than one. Seems worthwhile to keep it. Otherwise it seems fairly clear. I just hacked up a quick attempt; we'll see if it validates. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7450#comment:16> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #229: Integer overflow in array allocation
by GHC 23 Jun '15

23 Jun '15

#229: Integer overflow in array allocation -------------------------------------+------------------------------------- Reporter: josefs | Owner: rwbarton Type: bug | Status: new Priority: normal | Milestone: 7.12.1 Component: Core Libraries | Version: 7.9 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: None/Unknown | Unknown/Multiple Blocked By: | Test Case: Related Tickets: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- Changes (by gershomb): * cc: gershomb (added) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/229#comment:19> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)?
by GHC 23 Jun '15

23 Jun '15

#7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)? -------------------------------------+------------------------------------- Reporter: iustin | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.12.1 Component: Compiler | Version: 7.6.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple performance bug | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Revisions: -------------------------------------+------------------------------------- Comment (by simonpj): Ben, good work. I think you said that the culprit is `consDictBind`. Here are some ideas: * `snocDictBinds is only ever called with a singleton argument. Nuke it in favour of `snocDictBind`. * The calls to `snocDictBind` in `specBind` have just called `flattenDictBinds` which stupidly throws away all the free varaible information cached in the `DictBind`. Better to keep it! To achieve that * `snocDictBind` should take a `DictBind` not a `CoreBind` * `flattenDictBinds` should return a `DictBind` not a `CoreBind`. This is the place we are discarding the free var info. Instead, keep it! Does that make sense? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7450#comment:15> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)?
by GHC 23 Jun '15

23 Jun '15

#7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)? -------------------------------------+------------------------------------- Reporter: iustin | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.12.1 Component: Compiler | Version: 7.6.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple performance bug | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Revisions: -------------------------------------+------------------------------------- Comment (by bgamari): I think I finally found it (after struggling for far too long wondering why the profiles were so useless, eventually finding that I misspelled `OPTIONS_GHC`). `Specialise(pair_fvs)` shows up very strongly in the profile. Still trying to reason through what is happening. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7450#comment:14> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #10403: GHC crashes on a partial type signature
by GHC 23 Jun '15

23 Jun '15

#10403: GHC crashes on a partial type signature -------------------------------------+------------------------------------- Reporter: | Owner: Artyom.Kazak | Status: new Type: bug | Milestone: Priority: normal | Version: 7.10.1 Component: Compiler | Operating System: Linux Keywords: | Type of failure: Compile-time Architecture: x86_64 | crash (amd64) | Blocked By: Test Case: | Related Tickets: Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- The following code crashes GHC 7.10.1 as is, but doesn't result in a crash if the signature is specified completely or isn't specified at all (inferred): {{{#!hs {-# LANGUAGE PartialTypeSignatures #-} module Main where data I a = I a instance Functor I where fmap f (I a) = I (f a) newtype B t a = B a instance Functor (B t) where fmap f (B a) = B (f a) newtype H f = H (f ()) app :: H (B t) app = h (H . I) (B ()) h :: _ --h :: Functor m => (a -> b) -> m a -> H m h f b = (H . fmap (const ())) (fmap f b) }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10403> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 18

Re: [GHC] #7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)?
by GHC 23 Jun '15

23 Jun '15

#7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)? -------------------------------------+------------------------------------- Reporter: iustin | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.12.1 Component: Compiler | Version: 7.6.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple performance bug | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Revisions: -------------------------------------+------------------------------------- Comment (by bgamari): Here is GHC 7.11 output annotated with relative timestamps for, * with specialisation: https://github.com/bgamari/ghc-T7450/blob/24da000aae23eb8cbd4cec52015c64fa9… /with-spec.log * without specialisation: https://github.com/bgamari/ghc-T7450/blob/24da000aae23eb8cbd4cec52015c64fa9… /no-spec.log -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7450#comment:13> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)?
by GHC 23 Jun '15

23 Jun '15

#7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)? -------------------------------------+------------------------------------- Reporter: iustin | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.12.1 Component: Compiler | Version: 7.6.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple performance bug | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Revisions: -------------------------------------+------------------------------------- Comment (by bgamari): Ahh, yes, sorry, the above repository is at this point really just a couple of scripts for characterising the problem. `-dshow-passes` points at the specializer being much of the issue. Take the case of a 1000 constructor ADT. With GHC from `master`, the compilation takes roughly 75 seconds on my laptop. Of this it seems roughly 20 seconds is spent in the specialiser, and another 20 seconds are spent in the assembler. Otherwise the rest of the time is split between various simplifier iterations, with no single iteration taking more than a couple of seconds. If I disable specialisation the compilation time drops to 52 seconds, with the largest portion of this being a 7 second simplifier iteration. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7450#comment:12> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #10556: Add parsePattern entry point
by GHC 23 Jun '15

23 Jun '15

#10556: Add parsePattern entry point -------------------------------------+------------------------------------- Reporter: mpickering | Owner: Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.10.1 Component: Compiler | Operating System: Unknown/Multiple Keywords: | Type of failure: None/Unknown Architecture: | Blocked By: Unknown/Multiple | Related Tickets: Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- The parser currently exposes several entry points but not one which parses patterns. It would be useful for me if the pattern parser was also exposed. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10556> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #10507: Looping with type level naturals
by GHC 23 Jun '15

23 Jun '15

#10507: Looping with type level naturals -------------------------------------+------------------------------------- Reporter: jhendrix | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.1 Keywords: | Operating System: MacOS X Architecture: x86_64 | Type of failure: GHC rejects (amd64) | valid program Test Case: | Blocked By: Blocking: | Related Tickets: Differential Revisions: | -------------------------------------+------------------------------------- When compiling the code below, I run into an error with GHC 7.10.1: {{{#!hs {-# LANGUAGE DataKinds #-} {-# LANGUAGE EmptyDataDecls #-} {-# LANGUAGE ExplicitNamespaces #-} {-# LANGUAGE GADTs #-} {-# LANGUAGE KindSignatures #-} {-# LANGUAGE TypeOperators #-} module Loop ( ) where import Data.Type.Equality ( (:~:)(Refl) ) import Prelude (Maybe(..), undefined) import GHC.TypeLits ( Nat, type (<=)) data V (n::Nat) testEq :: (m <= n) => V m -> V n -> Maybe (m :~: n) testEq = undefined uext :: (1 <= m, m <= n) => V m -> V n -> V n uext e w = case testEq e w of Just Refl -> e }}} The error I get in GHC 7.10.1 is: {{{#!hs % ghc -c Loop.hs Loop.hs:21:9: Context reduction stack overflow; size = 102 Use -fcontext-stack=N to increase stack size to N (1 GHC.TypeLits.<=? n) ~ 'GHC.Types.True }}} GHC 7.8.4 compiles this without errors. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10507> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5