February 2015 - ghc-tickets

[GHC] #8710: Overlapping patterns warning misplaced
by GHC 27 Feb '16

27 Feb '16

#8710: Overlapping patterns warning misplaced -------------------------+------------------------------------------------- Reporter: | Owner: goldfire | Status: new Type: bug | Milestone: Priority: low | Version: 7.7 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: | -------------------------+------------------------------------------------- Consider {{{ {-# OPTIONS_GHC -fwarn-overlapping-patterns #-} len [] = 0 len (_:t) = 1 + len t len _ = error "urk" }}} GHC reports the problem with this definition at the first line of it. I think it would be more helpful to have it at the last line, which is the pattern that will never match. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8710> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 22

[GHC] #9917: ddump-llvm runs opt/llc even when -fllvm isnt set
by GHC 27 Feb '16

27 Feb '16

#9917: ddump-llvm runs opt/llc even when -fllvm isnt set -------------------------------------+------------------------------------- Reporter: carter | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.8.3 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Blocked By: | None/Unknown Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- to reproduce put {{{ main= putStrLn "hello world" }}} in hello.hs and then invoke `ghc hello.hs -O2 -v3 -fforce-recomp -ddump-llvm` then at some point you'll see something like {{{ *** LLVM Optimiser: opt-3.4 /var/folders/py/wgp_hj9d2rl3cx48yym_ynj00000gn/T/ghc88460_0/ghc88460_2.ll -o /var/folders/py/wgp_hj9d2rl3cx48yym_ynj00000gn/T/ghc88460_0/ghc88460_4.bc -O2 '--enable-tbaa=true' *** LLVM Compiler: llc-3.4 -O2 '-relocation-model=pic' /var/folders/py/wgp_hj9d2rl3cx48yym_ynj00000gn/T/ghc88460_0/ghc88460_4.bc -o /var/folders/py/wgp_hj9d2rl3cx48yym_ynj00000gn/T/ghc88460_0/ghc88460_5.lm_s '--enable-tbaa=true' '-mattr=+sse2' *** LLVM Mangler: }}} in the output -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9917> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #8176: Language extensions not registered
by GHC 20 Feb '16

20 Feb '16

#8176: Language extensions not registered --------------------------------+------------------------------------------ Reporter: duncan | Owner: thoughtpolice Type: bug | Status: new Priority: normal | Milestone: 7.8.1 Component: Compiler | Version: 7.7 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: tests/driver/T4437.hs Difficulty: Unknown | Blocking: Blocked By: | Related Tickets: | --------------------------------+------------------------------------------ tests/driver/T4437.hs does not seem to be being used properly. That test is there to make sure that ghc developers do not forget to register their new extensions with Cabal. But it looks like it's just being silenced and they are being forgotten again. Recall, the protocol is that all Haskell implementations can register their extensions in the central location (which is Language.Haskell.Extensions in the Cabal package) when they are happy that they are ready for public consumption. This matters to users because extensions that are not ready for public consumption (ie not registered) cannot be used in packages uploaded to hackage. Currently `tests/driver/T4437.hs` contains: {{{ expectedGhcOnlyExtensions = [ "AllowAmbiguousTypes", "RelaxedLayout", "AlternativeLayoutRule", "AlternativeLayoutRuleTransitional", "ExplicitNamespaces", "TypeHoles", "OverloadedLists", "EmptyCase", "AutoDeriveTypeable", "NegativeLiterals", "NullaryTypeClasses"] }}} This looks like a much bigger list than it should be. Surely not all of these extensions are private to ghc or are still under development and not to be used? But if they are listed here, then the test doesn't complain about them. And the whole point of this test is to stop ghc devs from forgetting to register them. I think we need to change the test so that we distinguish between extensions that are really not ready, and ones that are on-track to be public. It should be a release blocker for the "on-track to be public" ones to still not be registered. The test should have comments giving this guidance for ghc contributors. I'm not quite sure how we adjust the test to make that happen (ie different behaviour for RC-build vs normal hacking validate). -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8176> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 17

[GHC] #9993: PostfixOperators doesn't work for types
by GHC 19 Feb '16

19 Feb '16

#9993: PostfixOperators doesn't work for types -------------------------------------+------------------------------------- Reporter: shachaf | Owner: Type: feature | Status: new request | Milestone: Priority: normal | Version: 7.8.4 Component: Compiler | Operating System: Unknown/Multiple Keywords: | Type of failure: None/Unknown Architecture: | Blocked By: Unknown/Multiple | Related Tickets: Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- `PostfixOperators` is a useful extension on the value level that defines `(x OP)` to mean `((OP) x)` for an infix operator `OP`, rather than `(\y -> (OP) x y)`. This could be even more useful on the type level, where lambdas are unavailable (so sections don't exist at all) but postfix operators could still exist. This is already a common pseudocode idiom in some cases, e.g. `(r ->)` instead of `((->) r)`. Is there a particular syntactic reason to disallow it? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9993> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #9518: Improve error message for unacceptable role annotations
by GHC 19 Feb '16

19 Feb '16

#9518: Improve error message for unacceptable role annotations -------------------------------------+------------------------------------- Reporter: dmcclean | Owner: Type: feature request | Status: new Priority: low | Milestone: Component: Compiler | Version: 7.8.3 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Blocked By: | None/Unknown Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- It would be nice if this message: {{{ Role mismatch on variable a: Annotation says representational but role nominal is required while checking a role annotation for `Point' }}} Could be extended to say where the requirement arises. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9518> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #8981: ghc-pkg complains about missing haddock interface files
by GHC 19 Feb '16

19 Feb '16

#8981: ghc-pkg complains about missing haddock interface files --------------------------+------------------------------------------------ Reporter: | Owner: thoughtpolice | Status: new Type: bug | Milestone: 7.8.2 Priority: normal | Version: 7.8.1 Component: | Operating System: MacOS X Compiler | Type of failure: Incorrect result at runtime Keywords: ghc- | Test Case: pkg | Blocking: Architecture: x86_64 | (amd64) | Difficulty: | Unknown | Blocked By: | Related Tickets: | --------------------------+------------------------------------------------ As Conal reported on the mailing list[1], `ghc-pkg check` on Mavericks allegedly returns: {{{ bash-3.2$ ghc-pkg check Warning: haddock-interfaces: /Users/conal/.cabal/share/doc/x86_64-osx- ghc-7.8.1/uniplate-1.6.12/html/uniplate.haddock doesn't exist or isn't a file Warning: haddock-interfaces: /Users/conal/.cabal/share/doc/x86_64-osx- ghc-7.8.1/polyparse-1.9/html/polyparse.haddock doesn't exist or isn't a file Warning: haddock-interfaces: /Users/conal/.cabal/share/doc/x86_64-osx- ghc-7.8.1/ghc-syb-utils-0.2.1.2/html/ghc-syb-utils.haddock doesn't exist or isn't a file Warning: haddock-interfaces: /Users/conal/.cabal/share/doc/x86_64-osx- ghc-7.8.1/constraints-0.3.5/html/constraints.haddock doesn't exist or isn't a file Warning: haddock-html: /Users/conal/.cabal/share/doc/x86_64-osx- ghc-7.8.1/constraints-0.3.5/html doesn't exist or isn't a directory Warning: haddock-interfaces: /Users/conal/.cabal/share/doc/x86_64-osx- ghc-7.8.1/newtype-0.2/html/newtype.haddock doesn't exist or isn't a file Warning: haddock-html: /Users/conal/.cabal/share/doc/x86_64-osx- ghc-7.8.1/newtype-0.2/html doesn't exist or isn't a directory }}} It's not fatal, but makes the output much more annoying. I figured this would have been caught by validate or somesuch, but apparently not. Marking for 7.8.2. I'm looking into this soon. [1]http://www.haskell.org/pipermail/glasgow-haskell- users/2014-April/024846.html -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8981> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 6

[GHC] #9611: Suggest the cause of "No instance" errors involving multiple versions of a package
by GHC 18 Feb '16

18 Feb '16

#9611: Suggest the cause of "No instance" errors involving multiple versions of a package -------------------------------------+------------------------------------- Reporter: rwbarton | Owner: Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.8.3 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Blocked By: | None/Unknown Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- Suppose I have packages A-1.0, A-1.1, and B-1.0 built against A-1.0, and A (both versions) defines a type T which B defines to be an instance of some class C. If I load A-1.1 and B-1.0 into ghci and try to use the C T instance I will get a "No instance for C T" error, because T refers to A-1.1:T and B only defines an instance for A-1.0:T. However, if I don't realize this is what's going on, the error will be quite confusing as B appears to define an instance C T. Or, A could define the class C, rather than the type T, with the same result. This is quite common nowadays because there is a newer version of transformers (0.4.1.0) than the one which ships with GHC (0.3.0.0) and it's easy to end up installing mtl or lens against 0.3.0.0, and then some other package which pulls in 0.4.1.0. It would be nice to give a hint about what is going on, like {{{ No instance for (MonadIO X) arising from a use of ‘liftIO’ In the expression: ... ... Note: there is an instance transformers-0.3.0.0:MonadIO X }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9611> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 8

[GHC] #9648: ghci fails to reload after deleting/creating an imported module
by GHC 17 Feb '16

17 Feb '16

#9648: ghci fails to reload after deleting/creating an imported module -------------------------------------+------------------------------------- Reporter: NeilMitchell | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: GHCi | Version: 7.8.3 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Blocked By: | None/Unknown Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- If you start GHCi with a 2 module program, delete the child module, reload, you get an error (as expected). If you recreate the child module and reload, it reloads the dependent module but erroneously claims that parent can't find the child. Note that if you rename the child to something else, then rename it back, then the test works - so it seems the reappearance and the file stamp confuses ghci. Spotted while working on ghcid: https://github.com/ndmitchell/ghcid {{{ $ echo module Util where > Util.hs $ echo import Util; main = print 1 > Main.hs $ ghci Main.hs GHCi, version 7.8.3: http://www.haskell.org/ghc/ :? for help Loading package ghc-prim ... linking ... done. Loading package integer-gmp ... linking ... done. Loading package base ... linking ... done. [1 of 2] Compiling Util ( Util.hs, interpreted ) [2 of 2] Compiling Main ( Main.hs, interpreted ) Ok, modules loaded: Util, Main. *Main> :!del Util.hs *Main> :r Main.hs:1:8: Could not find module `Util' It is a member of the hidden package `ghc-7.8.3'. Use -v to see a list of the files searched for. Failed, modules loaded: Util, Main. *Main> :!echo module Util where > Util.hs *Main> :r [1 of 2] Compiling Util ( Util.hs, interpreted ) [2 of 2] Compiling Main ( Main.hs, interpreted ) [Util changed] Main.hs:1:1: Failed to load interface for `Util' It is a member of the hidden package `ghc-7.8.3'. Use -v to see a list of the files searched for. Failed, modules loaded: Util. }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9648> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #9041: NCG generates slow loop code
by GHC 17 Feb '16

17 Feb '16

#9041: NCG generates slow loop code -------------------------+------------------------------------------------- Reporter: nh2 | Owner: Type: bug | Status: new Priority: | Milestone: normal | Version: 7.6.3 Component: | Operating System: Unknown/Multiple Compiler (NCG) | Type of failure: Compile-time performance bug Keywords: | Test Case: Architecture: | Blocking: Unknown/Multiple | Difficulty: | Unknown | Blocked By: | Related Tickets: | -------------------------+------------------------------------------------- In http://stackoverflow.com/a/23322255/263061 we compile the code {{{ main :: IO () main = do loop (maxBound :: Word32) $ \i -> do when (i `rem` 100000000 == 0) $ putStrLn "foo" }}} and compare the assembly generated by the NCG and `-fllvm`. The LLVM code is 10 times faster. While the LLVM code looks great, NCG one generated looks quite unfortunate and messy, with a lot of unnecessary `mov`ing around. I am by no means an expert in this, but some of it looks like low hanging potential improvements. This is the '''LLVM code''': {{{ Main_zdwa_info: .LBB1_1: movl $4294967295, %esi /* load the loop bound */ movabsq $-6067343680855748867, %rdi /* load a magic number for the modulus */ jmp .LBB1_2 .LBB1_4: incl %ecx /* loop core start */ .LBB1_2: cmpq %rsi, %rcx je .LBB1_6 /* check loop bound */ /* do the modulus with two multiplications, a shift and a magic number */ /* note : gcc does the same reduction */ movq %rcx, %rax mulq %rdi shrq $26, %rdx imulq $100000000, %rdx, %rax cmpq %rax, %rcx jne .LBB1_4 /* loop core end */ /* Code omitted: print, then return to loop beginning */ .LBB1_6: }}} So the core of the loop is a nice and short: {{{ incl cmpq je movq [division replaced by magic multiplication] mulq shrq imulq cmpq jne }}} Now what '''NCG''' produces: {{{ Main_zdwa_info: /* loop core start */ .Lc3JD: leaq -16(%rbp),%rax /* stack check */ cmpq %r15,%rax jb .Lc3JO /* jump not taken in the normal case */ .Lc3JP: movl $4294967295,%eax /* issue: loading the bound on every iteration */ cmpq %rax,%r14 jne .Lc3JB .Lc3JC: /* Return from main. Code omitted */ .Lc3JB: /* test the index for modulus */ movl $100000000,%eax /* issue: unnecessary moves */ movq %rax,%rbx /* and loading the modulo arg on every iteration */ movq %r14,%rax xorq %rdx,%rdx /* shorter alternative to mov $0,%rdx */ divq %rbx /* issue: doing the division (llvm and gcc avoid this) */ testq %rdx,%rdx jne .Lc3JU /* This jump is usually taken. */ .Lc3JV: /* do the printing. Code omitted. Not relevant */ .Lc3JN: /* increment index and (I guess) restore registers messed up by the printing */ movq 8(%rbp),%rax /* This code is not very relevant because we don't almost never */ incq %rax movl %eax,%r14d addq $16,%rbp jmp Main_zdwa_info .Lc3JU: leaq 1(%r14),%rax /*issue: why not just increment r14? */ movl %eax,%r14d jmp Main_zdwa_info }}} So the core of this loop is {{{ movl [stack check] cmpq jne movl movq movq xorq divq testq jne leaq movl jmp }}} Some issues here: * the stack check is inside the loop * the loop limit is loaded every time inside the loop * the modulo arg is loaded every time inside the loop * we are shuffling around registers * no strength reduction (div is not replaced by cheaper magic number imul) * weird increment using leaq It would be great if somebody with codegen insight could comment on whether these are easy targets for improvement! -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9041> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #9637: Type level programming needs an error function
by GHC 17 Feb '16

17 Feb '16

#9637: Type level programming needs an error function -------------------------------------+------------------------------------- Reporter: augustss | Owner: Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.8.3 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Blocked By: | None/Unknown Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- When doing type level programming you often need a way to report errors. I suggest a closed type family called Error with special semantics. It is kinded like this Error :: Symbol -> k1 -> k2 If this function has to be reduced during constraint solving it will simply generate a type error with the given string and printing the type as extra information (this kind can be generalized a lot if we want to). I've found that having this function gives more accurate and earlier error messages. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9637> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 14