July 2013 - ghc-tickets

[GHC] #8061: Support for Complex Double in Foreign Function Interface
by GHC 06 Nov '13

06 Nov '13

#8061: Support for Complex Double in Foreign Function Interface ------------------------------------+------------------------------------- Reporter: dsamperi | Owner: Type: feature request | 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: | ------------------------------------+------------------------------------- Currently the FFI only permits BASIC return types like Double, Int, Bool, etc., but NOT Complex Double (only slightly non-basic!). This makes it impossible to import a Fortran function (not subroutine) that returns a complex value. Adding support for Complex Double would make it easier to interface to external libs like BLAS and LAPACK; without this support wrapper C libs must be used, and these are not standardized (the Haskell package hs-linear-algebra currently has some problems related to this). Thus adding Foreign support for Complex Double (not just through Ptr) would open up access to a large collection of external Fortran libs for the Haskell community. Currently I get the message "Unacceptable return type in FFI" when I try to return IO (ComplexType). -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8061> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #8047: Implement line-number %-substitution for prompt-string
by GHC 07 Oct '13

07 Oct '13

#8047: Implement line-number %-substitution for prompt-string ------------------------------------+------------------------------------- Reporter: hvr | Owner: Type: feature request | Status: new Priority: normal | Milestone: 7.8.1 Component: GHCi | 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: | ------------------------------------+------------------------------------- Currently, GHCi supports the following `%`-interpolations: > `:set prompt prompt` > > Sets the string to be used as the prompt in GHCi. Inside prompt, the sequence %s is replaced by the names of the modules currently in scope, and %% is replaced by %. If prompt starts with " then it is parsed as a Haskell String; otherwise it is treated as a literal string. The attached patch implements the `%l` variable, which gets replaced by the current line number of the shown prompt. The motivation being, that compiler messages refer to interactively entered bindings by the line number when they were defined, which are a bit hard to keep track of, e.g.: {{{ λ> :set prompt "\955:%l> " λ:3> let x = () x :: () λ:4> let x = ((),()) x :: ((), ()) λ:5> print xx <interactive>:5:7: Not in scope: ‛xx’ Perhaps you meant ‛x’ (line 4) λ:6> :info x x :: ((), ()) -- Defined at <interactive>:4:5 }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8047> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #8058: If .ghci is a symlink, permissions aren't read correctly
by GHC 24 Sep '13

24 Sep '13

#8058: If .ghci is a symlink, permissions aren't read correctly -------------------------------------------+------------------------------- Reporter: berdario | Owner: Type: bug | Status: new Priority: low | Milestone: Component: GHCi | Version: 7.6.2 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Easy (less than 1 hour) | Type of failure: Other Blocked By: | Test Case: Related Tickets: | Blocking: -------------------------------------------+------------------------------- dario@macbook ~> ls -l .ghci lrwxrwxrwx 1 dario dario 40 Jul 14 15:27 .ghci -> /home/dario/.dotfiles/dotfiles/ghci.conf dario@macbook ~> ls -l (readlink -f .ghci) -rw-r--r-- 1 dario dario 10 Jul 14 15:25 /home/dario/.dotfiles/dotfiles/ghci.conf dario@macbook ~> ghci GHCi, version 7.6.2: http://www.haskell.org/ghc/ :? for help Loading package ghc-prim ... linking ... done. Loading package integer-gmp ... linking ... done. Loading package base ... linking ... done. *** WARNING: /home/dario/.dotfiles/dotfiles is writable by someone else, IGNORING! Prelude> Obviously, /home/dario/.dotfiles/dotfiles isn't writable by someone else... Someone else could actually delete the symlink and maybe recreate it... but it couldn't inject malicious commands in the .ghci, unless the file pointed by the symlink was also writable by the attacker, in which case, checking the permissions of the target of the symlink would still prevent any wrongdoing -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8058> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 9

[GHC] #8051: GHCi: confusing `:set +m`/`:{` interaction & wrong line-numbers in multi-line input
by GHC 12 Sep '13

12 Sep '13

#8051: GHCi: confusing `:set +m`/`:{` interaction & wrong line-numbers in multi- line input ------------------------------------+------------------------------------- Reporter: hvr | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: GHCi | 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: | ------------------------------------+------------------------------------- As mentioned in ticket:8047#comment:1 GHCi exhibits a line-numbering offset bug when multi-line input is evaluated. Morever, when `:set +m` is active, `:{`-entered multi-input may trigger multi-line input continuation, e.g.: {{{ Prelude> :set +m Prelude> :{ Prelude| let x = () Prelude| y = () Prelude| :} Prelude| z = () Prelude| Prelude> }}} The attached patch fixes both issues, but does so in an admittedly hacky way... -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8051> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #8077: Make write_barrier and friends show up with KEEP_INLINES
by GHC 21 Aug '13

21 Aug '13

#8077: Make write_barrier and friends show up with KEEP_INLINES -------------------------------------+------------------------------------- Reporter: rrnewton | Owner: Type: bug | Status: new Priority: low | Milestone: Component: Runtime System | Version: 7.7 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: Compile-time Difficulty: Easy (less than 1 | crash hour) | Test Case: Blocked By: | Blocking: Related Tickets: | -------------------------------------+------------------------------------- The three barrier functions (write_barrier, store_load_barrier, load_load_barrier) are currently #defined macros when the non-threaded RTS is built. This means they are missed by the Inlines.c trick and can cause link errors in libraries that depend on them. The fix for this is obvious and this ticket is just a place to put the patch. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8077> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #8103: Segfault when passing unboxed Float# and Double# across modules
by GHC 15 Aug '13

15 Aug '13

#8103: Segfault when passing unboxed Float# and Double# across modules -----------------------------+---------------------------------- Reporter: jstolarek | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.7 Keywords: | Operating System: Linux Architecture: x86 | Type of failure: Runtime crash Difficulty: Unknown | Test Case: Blocked By: | Blocking: 6135 Related Tickets: | -----------------------------+---------------------------------- Consider this program consisting of two modules: {{{ {-# LANGUAGE MagicHash #-} module AddWraps where import GHC.Exts {-# NOINLINE foo #-} foo :: Double# -> Double# -> Double# foo a b = (a +## b) }}} {{{ {-# LANGUAGE MagicHash #-} module Main where import AddWraps float_text = case (0.0## `foo` 1.2##) of 0.0## -> "1" _ -> "0" main = putStrLn (float_text) }}} This program segfaults when compiled on i386 with HEAD: {{{ [t-jastol@cam-05-unx : ~] $HOME/master-i386/bin/ghc -fforce-recomp add- double-extern.hs [1 of 2] Compiling AddWraps ( AddWraps.hs, AddWraps.o ) [2 of 2] Compiling Main ( add-double-extern.hs, add-double- extern.o ) Linking add-double-extern ... [t-jastol@cam-05-unx : ~] ./add-double-extern Segmentation fault (core dumped) }}} The problem does not occur: * on x86_64 * on 7.6.3 * when optimizations are turned on * fot `Int#`, `Word#` and `Char#` - only `Float#` and `Double#` are affected Segfault happens on line 223 of `rts/ThreadPaused.c` (`switch (info->i.type)`), which probably means that the stack gets corrupted. It turns out that Cmm generated for the `foo` is the same in both cases: {{{ [section "data" { AddWraps.foo_closure: const AddWraps.foo_info; }, AddWraps.foo_slow() // [R1] { info_tbl: [] stack_info: arg_space: 20 updfr_space: Just 4 } {offset cfx: _rf9::P32 = R1; _B2::F64 = F64[Sp]; _B1::F64 = F64[Sp + 8]; D2 = _B1::F64; D1 = _B2::F64; R1 = _rf9::P32; Sp = Sp + 16; call AddWraps.foo_info(D2, D1, R1) args: 4, res: 0, upd: 4; } }, AddWraps.foo_entry() // [D2, D1] { info_tbl: [(cfB, label: AddWraps.foo_info rep:HeapRep static { Fun {arity: 2 fun_type: ArgGen [True, True, True, True]} })] stack_info: arg_space: 4 updfr_space: Just 4 } {offset cfB: _B1::F64 = D2; _B2::F64 = D1; goto cfD; cfD: _cfA::F64 = %MO_F_Add_W64(_B2::F64, _B1::F64); D1 = _cfA::F64; call (P32[Sp])(D1) args: 4, res: 0, upd: 4; } }] }}} The difference lies in how the calls are generated. When `foo` is in the same module as caller the generated call looks like this: {{{ cjY: I32[Sp - 8] = stg_bh_upd_frame_info; P32[Sp - 4] = Hp - 4; I32[Sp - 12] = ck0; D2 = 1.2 :: W64; D1 = 0.0 :: W64; Sp = Sp - 12; call Main.foo_info(D2, D1) returns to ck0, args: 4, res: 4, upd: 12; ck0: _sjs::F64 = D1; Hp = Hp + 12; if (Hp > HpLim) goto ckf; else goto ckc; }}} Whereas placing `foo` in separate module leads to this code: {{{ ck5: I32[Sp - 8] = stg_bh_upd_frame_info; P32[Sp - 4] = Hp - 4; I32[Sp - 12] = ck7; D1 = 0.0 :: W64; R1 = AddWraps.foo_closure; I32[Sp - 24] = stg_ap_d_info; F64[Sp - 20] = 1.0 :: W64; Sp = Sp - 24; call stg_ap_d_fast(D1, R1) returns to ck7, args: 16, res: 4, upd: 12; ck7: _sjP::F64 = D1; Hp = Hp + 12; if (Hp > HpLim) goto ckm; else goto ckj; }}} The second version causes a segfault. This currently blocks merging of #6135 patches. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8103> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 11

[GHC] #8106: Confusing parse error for missing do keyword
by GHC 01 Aug '13

01 Aug '13

#8106: Confusing parse error for missing do keyword -------------------------------------+-------------------------- Reporter: mhitza | Owner: Type: bug | Status: new Priority: low | Milestone: Component: Compiler (Parser) | Version: 7.6.3 Keywords: | Operating System: Linux Architecture: x86_64 (amd64) | Type of failure: Other Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | -------------------------------------+-------------------------- {{{ import Control.Monad (when) main = do when (True == True) $ x <- getLine print "y" }}} This piece of code is missing the `do` keyword before the block, yet the error message that is returned is: {{{ poc.hs:4:11: Parse error in pattern: True == True }}} Yet the next variant provides a bit more helpful information {{{ import Control.Monad (when) main = do when (True == True) $ print "x" print "y" }}} {{{ poc.hs:5:9: Couldn't match expected type `(a0 -> IO ()) -> [Char] -> m0 ()' with actual type `IO ()' The function `print' is applied to three arguments, but its type `[Char] -> IO ()' has only one In the second argument of `($)', namely `print "x" print "y"' In a stmt of a 'do' block: when (True == True) $ print "x" print "y" }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8106> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1

[GHC] #8105: unhelpful error message PolyKinds + GADTs
by GHC 31 Jul '13

31 Jul '13

#8105: unhelpful error message PolyKinds + GADTs -------------------------+------------------------------------------------- Reporter: wvv | Owner: Type: bug | Status: new Priority: | Milestone: normal | Version: 7.6.3 Component: | Operating System: Unknown/Multiple Compiler | Type of failure: Incorrect warning at Keywords: | compile-time Architecture: | Test Case: Unknown/Multiple | Blocking: Difficulty: | Unknown | Blocked By: | Related Tickets: | -------------------------+------------------------------------------------- For {{{ data Foo (a::k) = Foo a }}} we have next right error: {{{ Kind mis-match Expected kind `OpenKind', but `a' has kind `k' In the type `a' In the definition of data constructor `Foo' In the data declaration for `Foo' }}} but for same GADTs data we have just {{{ data Foo (a::k) where Foo a :: Foo a <interactive>:38:27: parse error on input `a' }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8105> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 1

Re: [GHC] #7895: Warning for literal overflow
by GHC 31 Jul '13

31 Jul '13

#7895: Warning for literal overflow -------------------------------------+------------------------------------ Reporter: acube | Owner: Type: feature request | Status: closed Priority: normal | Milestone: 7.8.1 Component: Compiler | Version: 7.6.3 Resolution: fixed | Keywords: Operating System: Unknown/Multiple | Architecture: Unknown/Multiple Type of failure: Other | Difficulty: Unknown Test Case: | Blocked By: Blocking: | Related Tickets: -------------------------------------+------------------------------------ Changes (by igloo): * status: new => closed * difficulty: => Unknown * resolution: => fixed * milestone: => 7.8.1 Comment: Thanks for the suggestion; now done. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7895#comment:3> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0

Re: [GHC] #7895: Warning for literal overflow
by GHC 31 Jul '13

31 Jul '13

#7895: Warning for literal overflow -------------------------------------+------------------------------------ Reporter: acube | Owner: Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.6.3 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Unknown/Multiple Type of failure: Other | Blocked By: Blocking: | Related Tickets: -------------------------------------+------------------------------------ Comment (by ian@…): commit 4e7eb3a0e32080acada355eae657e4e27465bc7e {{{ Author: Ian Lynagh <ian(a)well-typed.com> Date: Wed Jul 31 16:49:49 2013 +0100 Add a warning for overflowing literals; fixes #7895 compiler/deSugar/DsExpr.lhs | 39 +++++++++++++++++++++++++++++++++++++-- compiler/main/DynFlags.hs | 3 +++ docs/users_guide/using.xml | 13 +++++++++++++ 3 files changed, 53 insertions(+), 2 deletions(-) }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7895#comment:2> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 0