May 2015 - ghc-tickets

[GHC] #10317: Event manager: Multishot registrations only fire once
by GHC 01 Jun '15

01 Jun '15

#10317: Event manager: Multishot registrations only fire once -------------------------------------+------------------------------------- Reporter: bgamari | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.10.2 Component: Compiler | Version: 7.10.1 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: Incorrect result Unknown/Multiple | at runtime Test Case: | Blocked By: Blocking: | Related Tickets: Differential Revisions: | -------------------------------------+------------------------------------- Since D347 the event manager has had support for multishot event registration semantics, allowing the user to receive multiple events on an fd without requiring re-registration. Unfortunately the implementation drops the registration after it fires, resulting in only one callback invocation. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10317> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 9

[GHC] #10126: Test framework should not assume that GHC tools are in the same directory as GHC itself
by GHC 01 Jun '15

01 Jun '15

#10126: Test framework should not assume that GHC tools are in the same directory as GHC itself -------------------------------------+------------------------------------- Reporter: Rufflewind | Owner: Rufflewind Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.8.4 Component: Test Suite | Operating System: Unknown/Multiple Keywords: test | Type of failure: Other framework makefile | Blocked By: Architecture: | Related Tickets: Unknown/Multiple | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- The GHC test framework assumes that the GHC tools (`ghc-pkg`, `runghc`, `hsc2hs`, etc) are always in the same directory as `ghc` itself While this is probably a safe assumption for "in-tree" compilers, it's not necessarily true for "out-of-tree" compilers. For example, I currently have a thin wrapper of `ghc` in `$HOME/bin/ghc`, which causes the test framework to think that `ghc-pkg` is in `$HOME/bin/ghc-pkg`. I think the restriction should be relaxed as long as `TEST_HC` is not specified and an out-of-tree compiler is used. I suggest changing `testsuite/mk/boilerplate.mk` in the following way: {{{#!patch @@ -56,6 +56,7 @@ TEST_HC := $(STAGE2_GHC) endif else +implicit_compiler = YES IN_TREE_COMPILER = NO TEST_HC := $(shell which ghc) endif @@ -87,24 +88,30 @@ endif # containing spaces BIN_ROOT = $(shell dirname '$(TEST_HC)') +ifeq "$(implicit_compiler)" "YES" +find_tool = $(shell which $(1)) +else +find_tool = $(BIN_ROOT)/$(1) +endif + ifeq "$(GHC_PKG)" "" -GHC_PKG := $(BIN_ROOT)/ghc-pkg +GHC_PKG := $(call find_tool,ghc-pkg) endif ifeq "$(RUNGHC)" "" -RUNGHC := $(BIN_ROOT)/runghc +RUNGHC := $(call find_tool,runghc) endif ifeq "$(HSC2HS)" "" -HSC2HS := $(BIN_ROOT)/hsc2hs +HSC2HS := $(call find_tool,hsc2hs) endif ifeq "$(HP2PS_ABS)" "" -HP2PS_ABS := $(BIN_ROOT)/hp2ps +HP2PS_ABS := $(call find_tool,hp2ps) endif ifeq "$(HPC)" "" -HPC := $(BIN_ROOT)/hpc +HPC := $(call find_tool,hpc) endif $(eval $(call canonicaliseExecutable,TEST_HC)) }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10126> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 6

[GHC] #10360: GHC ignores command-line options if *.o and *.hi files exist
by GHC 31 May '15

31 May '15

#10360: GHC ignores command-line options if *.o and *.hi files exist -------------------------------------+------------------------------------- Reporter: asr | 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: | -------------------------------------+------------------------------------- For example, given the module {{{#!hs module Test where foo = putStrLn "Hello World" }}} if I compile using the command {{{ ghc -c Test.hs }}} and then I compile adding the `-Wall` flag {{{ ghc -c -Wall Test.hs }}} I would expect the warning {{{ Test.hs:3:1: Warning: Top-level binding with no type signature: foo :: IO () }}} but GHC produces the message {{{ compilation IS NOT required }}} That is, the `-Wall` flag was ignored because the `Test.o` and `Test.hi` exist. (I couldn't reproduce this behaviour using gcc version 4.6.3.) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10360> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #8292: linker_unload test doesn't work on Windows
by GHC 31 May '15

31 May '15

#8292: linker_unload test doesn't work on Windows ------------------------------------+---------------------------------- Reporter: ezyang | Owner: simonmar Type: bug | Status: new Priority: normal | Milestone: Component: Test Suite | Version: 7.7 Keywords: | Operating System: Windows Architecture: Unknown/Multiple | Type of failure: None/Unknown Difficulty: Unknown | Test Case: linker_unload Blocked By: | Blocking: Related Tickets: | ------------------------------------+---------------------------------- I fixed up the major issues but I am getting: {{{ "/c/Users/ezyang/Dev/ghc-dyn/inplace/bin/ghc-stage2.exe" -fforce-recomp -dcore-lint -dcmm-lint -dno-debug-output -no-user-package-db -rtsopts -c Test.hs -v0 # -rtsopts causes a warning "/c/Users/ezyang/Dev/ghc-dyn/inplace/bin/ghc-stage2.exe" -fforce-recomp -dcore-lint -dcmm-lint -dno-debug-output -no-user-package-db linker_unload.c -o linker_unload -no-hs-main -optc-Werror ./linker_unload \Users\ezyang\Dev\ghc-dyn\libraries\base\dist- install\build/libHSbase-4.7.0.0.a \Users\ezyang\Dev\ghc-dyn\libraries\ghc-prim\dist-install\build/libHSghc- prim-0.3.1.0.a \Users \ezyang\Dev\ghc-dyn\libraries\integer-gmp\dist-install\build/libHSinteger- gmp-0.5.1.0.a linker_unload.exe: \Users\ezyang\Dev\ghc-dyn\libraries\base\dist- install\build/libHSbase-4.7.0. 0.a: unknown symbol `_select' }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8292> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 6

[GHC] #9930: By default, a source code would get overwritten by executable on *nix
by GHC 31 May '15

31 May '15

#9930: By default, a source code would get overwritten by executable on *nix -------------------------------------+------------------------------------- Reporter: Rufflewind | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.8.4 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Other Blocked By: | Test Case: Related Tickets: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- If I have a source code without an extension, say `foo`, and I forget to specify the output file via `-o`, GHC will overwrite it with the executable itself after compilation. This occurs only on *nix-like operating systems. Presumably something similar could happen on Windows – fortunately, those who use `.exe` for source code are probably among the minority. Minimum example: {{{#!sh #!/bin/sh echo >foo 'main=undefined' ghc -x hs foo less foo # 'foo' now contains binary data }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9930> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 8

[GHC] #10465: Make listArray non-strict in structure of argument list
by GHC 30 May '15

30 May '15

#10465: Make listArray non-strict in structure of argument list -------------------------------------+------------------------------------- Reporter: atze | Owner: ekmett Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.10.1 Component: Core | Operating System: Unknown/Multiple Libraries | Type of failure: Runtime crash Keywords: laziness | Blocked By: array | Related Tickets: Architecture: | Unknown/Multiple | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- There is a subtle caveat when using Data.Array.IArray.listArray, which could be easily avoided. The problem is that listArray strict in the structure of the list, and hence if this structure depends on the result of listArray, then the result is undefined. However, listArray does not have to be strict in the structure of the list. As an example, consider computing the "failure function" for the Knuth- Morris-Pratt algorithm: {{{#!hs failureFunc :: Eq a => Array Int a -> Array Int Int failureFunc string = result where result = listArray (bounds string) (-1 : 0 : getMatch 2 0) getMatch :: Int -> Int -> [Int] getMatch p matchPos | matchPos < 0 = 0 : getMatch (p+1) 0 | string ! (p - 1) == string ! matchPos = matchPos + 1 : getMatch (p+1) (matchPos+1) | otherwise = getMatch p (result ! matchPos) -- use result! }}} This seems reasonable, we just use the result of elements < i to construct element i. However, it does not work: {{{#!hs Main> elems $ failureFunc (listArray (0,23) "participate in parachute") *** Exception: <<loop>> }}} The problem is that listArray is equivalent to: {{{#!hs listArray b l = array b (zip (range b) l) }}} (Recall that array is strict in the structure of the given list, and in the first elements of the tuples in this list, but not in the elements.) The structure of the list l=(-1 : 0 : getMatch 2 0) depends on the observing elements of the array (result ! matchPos). The structure of (zip (range b) l) is strict in the structure of l, and hence failure loops. Proposed solution: redefine listArray as equivalent to: {{{#!hs listArray b l = array b (zipLazyRight (range b) l) zipLazyRight :: [a] -> [b] -> [(a,b)] zipLazyRight [] _ = [] zipLazyRight (h:t) l = (h,head l): zipLazyRight t (tail l) }}} The difference is that listArray is now non-strict in the structure of the argument list, because zipLazyRight is non-strict in the right argument. Using this definition, failureFunc does not fail: {{{#!hs Main> elems $ failureFuncArr (toArr "participate in parachute") [-1,0,0,0,0,0,0,0,1,2,0,0,0,0,0,0,1,2,3,0,0,0,0,0] }}} This does not change the semantics uses of listArray which were already defined: it just makes more uses of listArray defined. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10465> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1

[GHC] #10261: Don't build runghc if we don't have GhcWithInterpreter
by GHC 30 May '15

30 May '15

#10261: Don't build runghc if we don't have GhcWithInterpreter -------------------------------------+------------------------------------- Reporter: rwbarton | Owner: Type: bug | Status: new Priority: low | Milestone: 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: | -------------------------------------+------------------------------------- On configurations with GhcWithInterpreter=NO, we should not build runghc (since it will never work), and especially not install runghc/runhaskell scripts/symlinks when the system might already have another working runhaskell (e.g., Hugs). -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10261> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

Re: [GHC] #6037: Compile-time crash with sources with non-representable unicode characters
by GHC 29 May '15

29 May '15

#6037: Compile-time crash with sources with non-representable unicode characters -------------------------------------+------------------------------------- Reporter: akio | Owner: Type: bug | Status: new Priority: normal | Milestone: 7.12.1 Component: Compiler | Version: 7.4.1 Resolution: | Keywords: Operating System: Linux | Architecture: Type of failure: Compile-time | Unknown/Multiple crash | Test Case: T6037 Blocked By: | Blocking: Related Tickets: | Differential Revisions: -------------------------------------+------------------------------------- Changes (by RyanGlScott): * cc: RyanGlScott (added) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/6037#comment:9> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #7695: Hang when locale-archive and gconv-modules are not there
by GHC 28 May '15

28 May '15

#7695: Hang when locale-archive and gconv-modules are not there -------------------------------------+------------------------------------- Reporter: hpd | Owner: Type: bug | Status: closed Priority: highest | Milestone: 7.10.2 Component: None | Version: 7.8.1 Resolution: fixed | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Runtime crash | Unknown/Multiple Blocked By: | Test Case: Related Tickets: #8977, #10298 | Blocking: | Differential Revisions: Phab:D898 -------------------------------------+------------------------------------- Changes (by thoughtpolice): * status: merge => closed * resolution: => fixed Comment: Merged to `ghc-7.10`. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7695#comment:22> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

[GHC] #10399: ApiAnnotations tweaks
by GHC 28 May '15

28 May '15

#10399: ApiAnnotations tweaks -------------------------------------+------------------------------------- Reporter: alanz | Owner: alanz Type: bug | Status: new Priority: normal | Milestone: 7.10.2 Component: Compiler | Version: 7.10.1 Keywords: | Operating System: Unknown/Multiple ApiAnnotations | Type of failure: None/Unknown Architecture: | Blocked By: Unknown/Multiple | Related Tickets: Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- A collection of minor updates for the API Annotations. 1. The annotations for the RdrNames `(+)` and `(-)` are disconnected in the following {{{#!hs import GHC.TypeLits (KnownNat, type (+), type (-)) }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10399> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 6