March 2018 - ghc-tickets

[GHC] #14891: Cabal bump broke ext-interp tests on Darwin
by GHC 26 Mar '18

26 Mar '18

#14891: Cabal bump broke ext-interp tests on Darwin -------------------------------------+------------------------------------- Reporter: bgamari | Owner: (none) Type: bug | Status: new Priority: highest | Milestone: 8.4.1 Component: Compiler | Version: 8.2.2 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: -------------------------------------+------------------------------------- Since a9f680f631e2 the testsuite has been failing with [[https://phabricator.haskell.org/harbormaster/build/42193|broken ext- interp tests]] on Darwin. I also see this on the ghc-8.4 branch. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14891> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 6

[GHC] #14911: Offer a way to augment call stacks
by GHC 26 Mar '18

26 Mar '18

#14911: Offer a way to augment call stacks -------------------------------------+------------------------------------- Reporter: dfeuer | Owner: (none) Type: feature | Status: new request | Priority: normal | Milestone: 8.6.1 Component: Core | Version: 8.4.1 Libraries | Keywords: CallStacks | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Sometimes, I may want to capture some information that can be reported when an error occurs. Suppose I have {{{#!hs f x y = g x (h y) g :: HasCallStack => ... g x y = ..... (error "whoopsy") .... }}} I may want `f` to record information about `y` that will be reported if `g` throws an exception. As far as I can tell, the only currently supported way to do this is horrible and limited: {{{#!hs f x y = unsafeDupablePerformIO $ catch (evaluate (g x (h y))) $ \e -> .... }}} I'd much rather have a function like {{{#!hs addMessage :: HasCallStack => String -> (HasCallStack => a) -> a }}} This would stick a string into the current call stack "frame" and call the argument. I imagine this can be implemented directly with the underlying implicit parameter, likely with a slight change to the `CallStack` representation. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14911> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 12

[GHC] #14660: Improve +RTS -t --machine-readable
by GHC 26 Mar '18

26 Mar '18

#14660: Improve +RTS -t --machine-readable -------------------------------------+------------------------------------- Reporter: duog | Owner: (none) Type: task | Status: new Priority: normal | Milestone: Component: Runtime | Version: 8.3 System | 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: -------------------------------------+------------------------------------- The output of +RTS -t --machine-readable is a shown Haskell list, so we can easily add fields. Let's add everything shown by +RTS -s . I've posted a rough patch that does that at Phab:D4303. If this is a good idea, I'd like to also add the following: * a data structure and parsing function to GHC.Stats in base. * The ghc version as a field in the --machine-readable output * The rts ways as a field in the --machine-readable output -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14660> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #14914: Only turn suitable targets into a fallthrough in CmmContFlowOpt.
by GHC 26 Mar '18

26 Mar '18

#14914: Only turn suitable targets into a fallthrough in CmmContFlowOpt. -------------------------------------+------------------------------------- Reporter: AndreasK | Owner: (none) Type: task | Status: new Priority: normal | Milestone: Component: Compiler | Version: (CodeGen) | 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: -------------------------------------+------------------------------------- We are inverting Branches in CmmContFlowOpt such that we get as many fall through paths as possible. However as it stands besides the conditionals themselves the only criterion used is the number of predecessors for a Block. This often (but not always) aligns with the actual possibility of turning branches into a fallthrough. We should at least look at the cost/benefit of tracking actual predecessors and see if that turns out as something worthwhile. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14914> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1

[GHC] #14967: Optimizer Casting Caf with nominal type
by GHC 26 Mar '18

26 Mar '18

#14967: Optimizer Casting Caf with nominal type -------------------------------------+------------------------------------- Reporter: etn | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.4.1 Keywords: coerce, | Operating System: Unknown/Multiple nominal | Architecture: | Type of failure: Incorrect result Unknown/Multiple | at runtime Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- When you make a function with a constant result, the optimizer will turn it into a cast of a CAF, even if doing so coerces a nominal parameter into a different one. Minimal Example program: {{{#!hs {-# LANGUAGE RoleAnnotations #-} import Debug.Trace (trace) type role Nom nominal data Nom a = Nom Int deriving Show class Gen g where gen :: g instance Gen (Nom a) where gen = trace "genD" $ Nom 0 main = print (gen :: Nom Int) >> print (gen :: Nom ()) >> print (gen :: Nom Char) }}} This program shows that only one value of type Nom is created and shared, even though doing so requires coercing a nominal role I discovered this while checking core for sharing after creating a constraint result caching mechanism. An IOref ended up being shared for multiple different constraint result holders -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14967> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 2

[GHC] #10536: Clear up how to turn off dynamic linking in build.mk
by GHC 26 Mar '18

26 Mar '18

#10536: Clear up how to turn off dynamic linking in build.mk -------------------------------------+------------------------------------- Reporter: thomie | Owner: Type: task | Status: new Priority: normal | Milestone: 7.12.1 Component: Build | Version: 7.10.1 System | Operating System: Unknown/Multiple Keywords: | Type of failure: None/Unknown Architecture: | Blocked By: Unknown/Multiple | Related Tickets: Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- See https://mail.haskell.org/pipermail/ghc-devs/2014-December/007725.html -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10536> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 8

[GHC] #14906: Release notes have wrong version of base package
by GHC 26 Mar '18

26 Mar '18

#14906: Release notes have wrong version of base package -------------------------------------+------------------------------------- Reporter: bgamari | Owner: (none) Type: bug | Status: new Priority: high | Milestone: 8.4.2 Component: Documentation | Version: 8.4.1 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: -------------------------------------+------------------------------------- The library version table introduced in the release notes in e4dc2cd51902a8cd83476f861cf52996e5adf157 claims that GHC 8.4.1 includes `base-2.1`. See https://downloads.haskell.org/~ghc/8.4.1/docs/html/users_guide/8.4.1-notes.… #included-libraries. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14906> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #14934: Repeated "impossible" go_axiom_rule error.
by GHC 26 Mar '18

26 Mar '18

#14934: Repeated "impossible" go_axiom_rule error. -------------------------------------+------------------------------------- Reporter: galen | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.4.1 Keywords: | Operating System: Linux Architecture: x86_64 | Type of failure: Compile-time (amd64) | crash or panic Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- I am getting the following error repeatedly: {{{ ghc: panic! (the 'impossible' happened) (GHC version 8.4.1 for x86_64-unknown-linux): go_axiom_rule Sub0R Call stack: CallStack (from HasCallStack): callStackDoc, called at compiler/utils/Outputable.hs:1150:37 in ghc:Outputable pprPanic, called at compiler/iface/TcIface.hs:1371:15 in ghc:TcIface }}} I got it on 8.2.1, and tried upgrading to 8.4.1 to see if it went away, but it did not. It occurs frequently when I do a `--make` and recompile a subset of modules. I can avoid it by force-recompiling all modules, although of course this is inefficient. Since I'm working with a codebase of thousands of lines and don't understand ghc's innards, I'm not sure where to start looking for the cause, to produce a minimal failing example. But it did start happening around the time I expanded my use of this module: https://github.com/agrafix/superrecord Since there are unsafe operations in there, it is possible it is doing something illicit, but it is hard to see how it would cause this. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14934> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 8

Re: [GHC] #5129: "evaluate" optimized away
by GHC 25 Mar '18

25 Mar '18

#5129: "evaluate" optimized away -------------------------------------+------------------------------------- Reporter: dons | Owner: (none) Type: bug | Status: closed Priority: normal | Milestone: 8.4.2 Component: Compiler | Version: 7.0.3 Resolution: fixed | Keywords: seq, evaluate Operating System: Unknown/Multiple | Architecture: Type of failure: Incorrect result | Unknown/Multiple at runtime | Test Case: Blocked By: | Blocking: Related Tickets: #13930 | Differential Rev(s): Phab:D615, Wiki Page: | Phab:D4514 -------------------------------------+------------------------------------- Changes (by bgamari): * status: merge => closed * resolution: => fixed Comment: Merged to `ghc-8.4`. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/5129#comment:38> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #11188: Confusing "parse error in pattern" for missing indentation.
by GHC 25 Mar '18

25 Mar '18

#11188: Confusing "parse error in pattern" for missing indentation. -------------------------------------+------------------------------------- Reporter: andreas.abel | Owner: Type: bug | Status: new Priority: high | Milestone: Component: Compiler | Version: 7.10.1 (Parser) | Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: Other Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- The following problem exists at least in ghc 7.4.2 through 7.10.1: {{{#!hs main = do putStrLn "Hello, lots of crap" $ do a <- return 3 -- The following line is mis-indented, the error is incomprehensible c <- do return 5 }}} Error: {{{ <file>:2:3: Parse error in pattern: putStrLn Possibly caused by a missing 'do'? }}} The hint about "missing do" exists from ghc 7.8, but is wrong in this case. Problematic about this bug is that the wrong indentation can be arbitrary many lines from where ghc reports the (totally incomprehensible) error. My original problem is to find the syntax error here: {{{#!hs -- | Type check a function clause. checkClause :: Type -> A.SpineClause -> TCM Clause checkClause t c(a)(A.Clause (A.SpineLHS i x aps withPats) rhs0 wh) = do unless (null withPats) $ do typeError $ UnexpectedWithPatterns withPats traceCall (CheckClause t c) $ do aps <- expandPatternSynonyms aps checkLeftHandSide (CheckPatternShadowing c) (Just x) aps t $ \ (LHSResult delta ps trhs perm) -> do -- Note that we might now be in irrelevant context, -- in case checkLeftHandSide walked over an irrelevant projection pattern. -- As we will be type-checking the @rhs@ in @delta@, but the final -- body should have bindings in the order of the pattern variables, -- we need to apply the permutation to the checked rhs @v@. let mkBody v = foldr (\ x t -> Bind $ Abs x t) b xs where b = Body $ applySubst (renamingR perm) v xs = [ stringToArgName $ "h" ++ show n | n <- [0..permRange perm - 1] ] -- introduce trailing implicits for checking the where decls TelV htel t0 <- telViewUpTo' (-1) (not . visible) $ unArg trhs (body, with) <- do let n = size htel addCtxTel htel $ checkWhere (size delta + n) wh $ -- for the body, we remove the implicits again escapeContext n $ handleRHS aps (unArgs trhs) rhs0 escapeContext (size delta) $ checkWithFunction with reportSDoc "tc.lhs.top" 10 $ escapeContext (size delta) $ vcat [ text "Clause before translation:" , nest 2 $ vcat [ text "delta =" <+> prettyTCM delta , text "perm =" <+> text (show perm) , text "ps =" <+> text (show ps) , text "body =" <+> text (show body) , text "body =" <+> prettyTCM body ] ] return $ Clause { clauseRange = getRange i , clauseTel = killRange delta , clausePerm = perm , namedClausePats = ps , clauseBody = body , clauseType = Just trhs } }}} {{{ src/full/Agda/TypeChecking/Rules/Def.hs:328:5: Parse error in pattern: checkLeftHandSide Possibly caused by a missing 'do'? }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11188> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 9