June 2013 - ghc-devs - Haskell.org

CPP macro to determine GHCi presence?
by Stephen Paul Weber 14 Jun '13

14 Jun '13

Trying to cross-build the vector package I get: > ghc-stage1: panic! (the 'impossible' happened) > (GHC version 7.7.20130123 for arm-unknown-nto-qnx): > Cant do annotations without GHCi > {Data/Vector/Fusion/Stream/Monadic.hs:104:19-33} > base:GHC.Exts.ForceSpecConstr{d r2Qv} This is because the source uses a feature that cannot be supported in cross-compilation. The problematic line, however, can easily be removed. It would be very useful to have a CPP macro defined by GHC that can be used to enable/disable these sorts of features. -- Stephen Paul Weber, @singpolyma See <http://singpolyma.net> for how I prefer to be contacted edition right joseph

1 0

spam on the wiki?
by Nicolas Frisby 13 Jun '13

13 Jun '13

I've yet to fully comprehend the build system, but I'm pretty sure these links are not related. http://hackage.haskell.org/trac/ghc/wiki/Building/Hacking?action=diff&versi…

3 6

exprType and existentials
by Andrew Farmer 13 Jun '13

13 Jun '13

Calling `exprType` on: case enumFromStepN Int $fNumInt (I# 1) (I# 1) x of wild Stream s1 ostep t ds1 -> Left s1 ((Int, (Int, Int)), s1) t gives: Either s1 ((Int, (Int, Int)), s1) Which is problematic, as the type variable 's1' is unbound outside of the alternative RHS! Unfolding the scrutinee and case-reducing gives: Left (Int, Int) ((Int, (Int, Int)), (Int, Int)) ((,) Int Int (I# 1) (case x of n1 I# ipv ? n1)) Calling `exprType` now gives: Either (Int, Int) ((Int, (Int, Int)), (Int, Int)) That is, `s1` was `(Int, Int)` in this case. Is it expected that `exprType` behaves this way? Or should I file a ticket? I realize the type of the scrutinee (in this case Stream Int) gives no indication as to the existential type... and I'm not sure how you would get it without making exprType do some kind of evaluation, but it would be nice if exprType at least failed if it doesn't have enough information... Andrew

2 1

Linking to changeset on wiki
by Geoffrey Mainland 12 Jun '13

12 Jun '13

What is the magical incantation for linking to a changeset in a bug report/on the trac wiki? i can't for the life of me figure it out. Thanks, Geoff

2 1

Build broken: urgent
by Simon Peyton-Jones 11 Jun '13

11 Jun '13

Ian This getRegister bug is killing my Windows builds too; indeed every 32-bit build is broken, and that is stalling a lot of people, now including me. Could you perhaps revert the patch (around 6 June) that caused this, until it's sorted out? Thanks Simon "inplace/bin/ghc-stage1.exe" -hisuf hi -osuf o -hcsuf hc -static -H32m -O -Werror -Wall -H64m -O0 -package-name ghc-prim-0.3.1.0 -hide-all-packages -i -ilibraries/ghc-prim/. -ilibraries/ghc-prim/dist-install/build -ilibraries/ghc-prim/dist-install/build/autogen -Ilibraries/ghc-prim/dist-install/build -Ilibraries/ghc-prim/dist-install/build/autogen -Ilibraries/ghc-prim/. -optP-include -optPlibraries/ghc-prim/dist-install/build/autogen/cabal_macros.h -package rts-1.0 -package-name ghc-prim -XHaskell98 -XCPP -XMagicHash -XForeignFunctionInterface -XUnliftedFFITypes -XUnboxedTuples -XEmptyDataDecls -XNoImplicitPrelude -O2 -O -dcore-lint -fno-warn-deprecated-flags -no-user-package-db -rtsopts -odir libraries/ghc-prim/dist-install/build -hidir libraries/ghc-prim/dist-install/build -stubdir libraries/ghc-prim/dist-install/build -dynamic-too -c libraries/ghc-prim/dist-install/build/GHC/PrimopWrappers.hs -o libraries/ghc-prim/dist-install/build/GHC/PrimopWrappers.o -dyno libraries/ghc-prim/dist-install/build/GHC/PrimopWrappers.dyn_o ghc-stage1.exe: panic! (the 'impossible' happened) (GHC version 7.7.20130611 for i386-unknown-mingw32): getRegister(x86) I64[Sp] Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug bash-3.1$

3 6

RE: builders
by Simon Peyton-Jones 11 Jun '13

11 Jun '13

Breakage on 32-bit Linux looks like this: ghc-stage1: panic! (the 'impossible' happened) (GHC version 7.7.20130610 for i386-unknown-linux): getRegister(x86) I64[Sp] Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug make[1]: *** [libraries/ghc-prim/dist-install/build/GHC/PrimopWrappers.o] Error 1 Breakage on 64-bit Linux started about 6 June, and looks like this: ../../libraries/array/tests T2120 [exit code non-0] (profasm,profthreaded) ../../libraries/array/tests array001 [exit code non-0] (profasm,profthreaded) ../../libraries/array/tests largeArray [exit code non-0] (profasm,profthreaded) ../../libraries/base/tests CPUTime001 [exit code non-0] (profasm,profthreaded) plus a million more From: Simon Peyton-Jones Sent: 11 June 2013 12:04 To: Ian Lynagh (ian(a)well-typed.com) Cc: ghc-devs(a)haskell.org Subject: builders Ian Something we didn't discuss yesterday: * 32-bit Linux build is broken (see nightly-build email that Geoff generates here at MSR) * are other buildbots working? * we are getting no snapshot builds * there are many test failures (with plain "make" in testsuite/) even on 64-bit Linux. Can you investigate? Thanks Simon Microsoft Research Limited (company number 03369488) is registered in England and Wales Registered office is at 21 Station Road, Cambridge, CB1 2FB

2 1

Validate failures
by Simon Peyton-Jones 11 Jun '13

11 Jun '13

Validation failures with "make" on 64-bit Linux are these. No prof or profasm tests were run; not sure why. But they are failing in droves Unexpected passes: driver T5313 (ghci,dyn) perf/should_run T149 (normal) Unexpected failures: ../../libraries/base/tests echo001 [bad exit code] (ghci) ../../libraries/base/tests/IO hClose002 [bad exit code] (ghci,threaded1,threaded2) ../../libraries/base/tests/IO hSetBuffering002 [bad exit code] (ghci) ../../libraries/base/tests/IO hSetBuffering004 [bad stdout or stderr] (ghci) ../../libraries/process/tests process007 [bad stderr] (threaded1,threaded2) ../../libraries/process/tests process007 [bad stdout or stderr] (ghci) ../../libraries/unix/tests/libposix posix002 [bad stderr] (threaded2) annotations/should_run annrun01 [bad exit code] (threaded2,optllvm) ffi/should_run T7170 [bad stdout or stderr] (ghci) ghc-api/apirecomp001 apirecomp001 [bad exit code] (normal) perf/space_leaks space_leak_001 [stat too good] (hpc,optasm,dyn,optllvm) programs/andy_cherry andy_cherry [bad exit code] (hpc,ghci) programs/jl_defaults jl_defaults [bad exit code] (ghci) programs/jq_readsPrec jq_readsPrec [bad exit code] (ghci) programs/jules_xref jules_xref [bad exit code] (ghci) programs/jules_xref2 jules_xref2 [bad exit code] (ghci) programs/launchbury launchbury [bad exit code] (ghci) programs/lex lex [bad exit code] (ghci) programs/rittri rittri [bad exit code] (ghci) quasiquotation/qq007 qq007 [exit code non-0] (normal,hpc,optasm,optllvm) quasiquotation/qq008 qq008 [exit code non-0] (normal,hpc,optasm,optllvm) Microsoft Research Limited (company number 03369488) is registered in England and Wales Registered office is at 21 Station Road, Cambridge, CB1 2FB

2 1

builders
by Simon Peyton-Jones 11 Jun '13

11 Jun '13

Ian Something we didn't discuss yesterday: * 32-bit Linux build is broken (see nightly-build email that Geoff generates here at MSR) * are other buildbots working? * we are getting no snapshot builds * there are many test failures (with plain "make" in testsuite/) even on 64-bit Linux. Can you investigate? Thanks Simon Microsoft Research Limited (company number 03369488) is registered in England and Wales Registered office is at 21 Station Road, Cambridge, CB1 2FB

1 0

development snapshot
by Richard Eisenberg 09 Jun '13

09 Jun '13

Hello all, I recently posted a bug report about cabal that's exposed by its interaction with GHC HEAD, and one of the cabal devs reasonably asked about a snapshot of HEAD somewhere. The link from http://www.haskell.org/ghc/download leads to an empty directory. Is there another place I can offer? Is there some script somewhere that has stopped doing its job? Thanks, Richard

2 1

Re: [commit: base] master: IO manager: Edit the timeout queue directly, rather than using an edit list (e843e73)
by Johan Tibell 08 Jun '13

08 Jun '13

Is this related to some bug? The edit list was there for a reason. :) On Jun 8, 2013 1:19 PM, "Ian Lynagh" <igloo(a)earth.li> wrote: > Repository : ssh://darcs.haskell.org//srv/darcs/packages/base > > On branch : master > > > https://github.com/ghc/packages-base/commit/e843e73690f828498f6e33bb89f47a5… > > >--------------------------------------------------------------- > > commit e843e73690f828498f6e33bb89f47a50c3ab2ac9 > Author: Ian Lynagh <ian(a)well-typed.com> > Date: Sat Jun 8 20:19:59 2013 +0100 > > IO manager: Edit the timeout queue directly, rather than using an edit > list > > Fixes #7653. > > >--------------------------------------------------------------- > > GHC/Event/TimerManager.hs | 61 > +++++++++++++++++++++----------------------- > 1 files changed, 29 insertions(+), 32 deletions(-) > > diff --git a/GHC/Event/TimerManager.hs b/GHC/Event/TimerManager.hs > index b581891..453f2eb 100644 > --- a/GHC/Event/TimerManager.hs > +++ b/GHC/Event/TimerManager.hs > @@ -39,7 +39,7 @@ module GHC.Event.TimerManager > > import Control.Exception (finally) > import Control.Monad ((=<<), liftM, sequence_, when) > -import Data.IORef (IORef, atomicModifyIORef, mkWeakIORef, newIORef, > readIORef, > +import Data.IORef (IORef, atomicModifyIORef, atomicModifyIORef', > mkWeakIORef, newIORef, readIORef, > writeIORef) > import Data.Maybe (Maybe(..)) > import Data.Monoid (mempty) > @@ -114,7 +114,7 @@ type TimeoutEdit = TimeoutQueue -> TimeoutQueue > -- | The event manager state. > data TimerManager = TimerManager > { emBackend :: !Backend > - , emTimeouts :: {-# UNPACK #-} !(IORef TimeoutEdit) > + , emTimeouts :: {-# UNPACK #-} !(IORef TimeoutQueue) > , emState :: {-# UNPACK #-} !(IORef State) > , emUniqueSource :: {-# UNPACK #-} !UniqueSource > , emControl :: {-# UNPACK #-} !Control > @@ -144,7 +144,7 @@ new = newWith =<< newDefaultBackend > > newWith :: Backend -> IO TimerManager > newWith be = do > - timeouts <- newIORef id > + timeouts <- newIORef Q.empty > ctrl <- newControl True > state <- newIORef Created > us <- newSource > @@ -192,38 +192,39 @@ loop mgr = do > Created -> (Running, s) > _ -> (s, s) > case state of > - Created -> go Q.empty `finally` cleanup mgr > + Created -> go `finally` cleanup mgr > Dying -> cleanup mgr > _ -> do cleanup mgr > error $ "GHC.Event.Manager.loop: state is already " ++ > show state > where > - go q = do (running, q') <- step mgr q > - when running $ go q' > + go = do running <- step mgr > + when running go > > -step :: TimerManager -> TimeoutQueue -> IO (Bool, TimeoutQueue) > -step mgr tq = do > - (timeout, q') <- mkTimeout tq > +step :: TimerManager -> IO Bool > +step mgr = do > + timeout <- mkTimeout > _ <- I.poll (emBackend mgr) (Just timeout) (handleControlEvent mgr) > state <- readIORef (emState mgr) > - state `seq` return (state == Running, q') > + state `seq` return (state == Running) > where > > -- | Call all expired timer callbacks and return the time to the > -- next timeout. > - mkTimeout :: TimeoutQueue -> IO (Timeout, TimeoutQueue) > - mkTimeout q = do > + mkTimeout :: IO Timeout > + mkTimeout = do > now <- getMonotonicTime > - applyEdits <- atomicModifyIORef (emTimeouts mgr) $ \f -> (id, f) > - let (expired, q'') = let q' = applyEdits q in q' `seq` Q.atMost now > q' > + (expired, timeout) <- atomicModifyIORef (emTimeouts mgr) $ \tq -> > + let (expired, tq') = Q.atMost now tq > + timeout = case Q.minView tq' of > + Nothing -> Forever > + Just (Q.E _ t _, _) -> > + -- This value will always be positive since the call > + -- to 'atMost' above removed any timeouts <= 'now' > + let t' = t - now in t' `seq` Timeout t' > + in (tq', (expired, timeout)) > sequence_ $ map Q.value expired > - let timeout = case Q.minView q'' of > - Nothing -> Forever > - Just (Q.E _ t _, _) -> > - -- This value will always be positive since the call > - -- to 'atMost' above removed any timeouts <= 'now' > - let t' = t - now in t' `seq` Timeout t' > - return (timeout, q'') > + return timeout > > -- | Wake up the event manager. > wakeManager :: TimerManager -> IO () > @@ -244,21 +245,14 @@ registerTimeout mgr us cb = do > now <- getMonotonicTime > let expTime = fromIntegral us / 1000000.0 + now > > - -- We intentionally do not evaluate the modified map to WHNF here. > - -- Instead, we leave a thunk inside the IORef and defer its > - -- evaluation until mkTimeout in the event loop. This is a > - -- workaround for a nasty IORef contention problem that causes the > - -- thread-delay benchmark to take 20 seconds instead of 0.2. > - atomicModifyIORef (emTimeouts mgr) $ \f -> > - let f' = (Q.insert key expTime cb) . f in (f', ()) > + editTimeouts mgr (Q.insert key expTime cb) > wakeManager mgr > return $ TK key > > -- | Unregister an active timeout. > unregisterTimeout :: TimerManager -> TimeoutKey -> IO () > unregisterTimeout mgr (TK key) = do > - atomicModifyIORef (emTimeouts mgr) $ \f -> > - let f' = (Q.delete key) . f in (f', ()) > + editTimeouts mgr (Q.delete key) > wakeManager mgr > > -- | Update an active timeout to fire in the given number of > @@ -268,6 +262,9 @@ updateTimeout mgr (TK key) us = do > now <- getMonotonicTime > let expTime = fromIntegral us / 1000000.0 + now > > - atomicModifyIORef (emTimeouts mgr) $ \f -> > - let f' = (Q.adjust (const expTime) key) . f in (f', ()) > + editTimeouts mgr (Q.adjust (const expTime) key) > wakeManager mgr > + > +editTimeouts :: TimerManager -> TimeoutEdit -> IO () > +editTimeouts mgr g = atomicModifyIORef' (emTimeouts mgr) $ \tq -> (g tq, > ()) > + > > > > _______________________________________________ > ghc-commits mailing list > ghc-commits(a)haskell.org > http://www.haskell.org/mailman/listinfo/ghc-commits >

2 1