February 2016 - ghc-tickets

[GHC] #8865: Cannot derive well-kinded instance of form ‘Category
by GHC 03 May '16

03 May '16

#8865: Cannot derive well-kinded instance of form ‘Category ----------------------------+---------------------------------------------- Reporter: | Owner: adinapoli | Status: new Type: bug | Milestone: 7.8.1 Priority: normal | Version: 7.8.1-rc2 Component: Compiler | Operating System: MacOS X Keywords: | Type of failure: GHC rejects valid program Architecture: x86_64 | Test Case: (amd64) | Blocking: Difficulty: Unknown | Blocked By: | Related Tickets: | ----------------------------+---------------------------------------------- Hello everyone, sorry if this was already reported somewhere. I'm playing with GHC 7.8.1 RC2 and I'm updating hsenv to support GHC 7.8. This is a real code snippet, which was compiling fine in GHC 7.6.x {{{ newtype ArgArrow a b = ArgArrow (StaticArrowT KnownArgs (Kleisli (ReaderT Args IO)) a b) deriving (Category, Arrow, ArrowChoice) }}} but that yields the following in GHC 7.8.1-RC2 {{{ Cannot derive well-kinded instance of form ‘Category (ArgArrow ...)’ Class ‘Category’ expects an argument of kind ‘k -> k -> *’ In the newtype declaration for ‘ArgArrow’ }}} This might be related to the new feature of GHC 7.8, namely "kind variables" (you get the idea, even if the name is not 100% accurate). To make the code compile I had to enable {{{ StandaloneDeriving }}} and write the following: {{{ newtype ArgArrow a b = ArgArrow (StaticArrowT KnownArgs (Kleisli (ReaderT Args IO)) a b) deriving (Arrow, ArrowChoice) deriving instance Category ArgArrow }}} Is this by design or is a genuine bug? Thanks! Alfredo -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8865> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 14

[GHC] #10266: Split base for Backpack
by GHC 03 May '16

03 May '16

#10266: Split base for Backpack -------------------------------------+------------------------------------- Reporter: ezyang | Owner: ezyang Type: task | Status: new Priority: normal | Milestone: Component: | Version: 7.11 libraries/base | Operating System: Unknown/Multiple Keywords: backpack | Type of failure: None/Unknown Architecture: | Blocked By: Unknown/Multiple | Related Tickets: Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- Let's resurrect #1338 and #1580. Our new concrete goal is to have users be able to replace 'base' with 'base-sig', which is some signature representing the modules from 'base' which we depend on. Since 'base' is a big, complicated package, the ability to specify subsets of 'base' is very useful: it can be used to define things like Haskell'98 base, base without any unsafe operations, etc. Right now, all wired in things live in ghc-prim (I believe). This is good, since we can't reasonably abstract wired in things away with a signature: GHC will always put them in some well-known place. However, this is not enough: there are many known-key/Rdr things the compiler knows about too, which need to be placed in specific package + module locations for GHC to be able to reference them; e.g. when deriving instances, etc. At the moment, most of these identifiers live in 'base', which is no good. So, here's the proposal: all non-rebindable GHC-known identifiers (as per PrelNames) should be moved to ghc-prim (or perhaps a new package? I don't really care either way) and reexported from the old packages which they were available from base. I'll try to work out in detail what moves, how to avoid orphan instances, etc. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10266> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #10520: RecordWildCards causes “is not a (visible) field of constructor” in ghci
by GHC 02 May '16

02 May '16

#10520: RecordWildCards causes “is not a (visible) field of constructor” in ghci -------------------------------------+------------------------------------- Reporter: ion1 | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: GHCi | Version: 7.10.1 Keywords: | Operating System: Linux RecordWildCards | Type of failure: GHC rejects Architecture: x86_64 | valid program (amd64) | Blocked By: Test Case: | Related Tickets: Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- {{{ % ghc -ignore-dot-ghci -e 'data Foo = Bar { baz :: Integer } deriving Show' -e 'Bar 42' -e '(Bar 42){ baz = 43 }' -e 'Bar { baz = 42 }' Bar {baz = 42} Bar {baz = 43} Bar {baz = 42} }}} {{{ % ghc -ignore-dot-ghci -XRecordWildCards -e 'data Foo = Bar { baz :: Integer } deriving Show' -e 'Bar 42' -e '(Bar 42){ baz = 43 }' -e 'Bar { baz = 42 }' Bar {baz = 42} Bar {baz = 43} <interactive>:1:7: ‘baz’ is not a (visible) field of constructor ‘Bar’ }}} {{{ % ghc --version The Glorious Glasgow Haskell Compilation System, version 7.10.1 }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10520> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 6

[GHC] #11545: Strictness signature blowup
by GHC 01 May '16

01 May '16

#11545: Strictness signature blowup -------------------------------------+------------------------------------- Reporter: jscholl | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.3 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: Compile-time Unknown/Multiple | performance bug Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- The following code takes a lot of memory (556MB) to compile: {{{#!hs data A = A A A A A A deriving (Eq) }}} More {{{A}}}s result in more memory consumption during compilation. Using {{{ghc -O A.hs -v -dverbose-core2core -c -ddump-simpl +RTS -s}}} it seems like the strictness signatures of (==) and (/=) blow up quite a bit. A less silly example could be the following: {{{#!hs data QuadTree a = QuadTree !Int [a] (QuadTree a) (QuadTree a) (QuadTree a) (QuadTree a) foldQuadTree :: (a -> b -> b) -> Int -> b -> QuadTree a -> b foldQuadTree f maxSize = go where go z (QuadTree size elems t1 t2 t3 t4) | size <= maxSize = foldr f z elems | otherwise = go (go (go (go z t4) t3) t2) t1 }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11545> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #9758: By default, testsuite should clean up after successful tests
by GHC 30 Apr '16

30 Apr '16

#9758: By default, testsuite should clean up after successful tests -------------------------------------+------------------------------------- Reporter: ezyang | Owner: Type: task | Status: new Priority: low | Milestone: Component: Test Suite | Version: 7.9 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Blocked By: | None/Unknown Related Tickets: | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- A fully built GHC compiler for development takes about 1G of space. However, running the testsuite balloons this amount to 10G. Why? Because we don't clear away any test files by default. Now, it's pretty useful to have these files if you're debugging a problem. But I think it's pretty unnecessary for successful tests. So I suggest cleaning out the files automatically if the test succeeds. Tradeoff: the testsuite will take longer to run since it is cleaning up after itself. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9758> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #9543: Testsuite driver: replace "extra_clean" by "git clean -X"
by GHC 26 Apr '16

26 Apr '16

#9543: Testsuite driver: replace "extra_clean" by "git clean -X" -------------------------------------+------------------------------------- Reporter: thomie | Owner: thomie Type: task | Status: new Priority: normal | Milestone: Component: Test Suite | Version: 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 adding a new test that creates extra files, one also has to do some bookkeeping: 1. Supply an `extra_clean` argument to the test function in `all.T`. 2. Add an entry in `testsuite/.gitignore` Step number 1 is often forgotten. Currently there are over 200 uncleaned files and directories left behind in the `testsuite`, after running `validate` and then `make clean` (checked with `git clean -X -d -n | wc -l`). I propose to replace the functionality that `extra_clean` provides, namely to remove these extra files when one calls `make clean`, by doing the equivalent of `git clean -X -d`: {{{ -X Remove only files ignored by Git. -d Remove untracked directories in addition to untracked files. ... }}} That is: use the information from the `.ignore` files to figure out which files to delete on `make clean`. The advantages of making this change are: 1. Less manual bookkeeping, less work, less code 2. A more thorough `make clean` Implementation detail: we can not just call `git clean` directly, since this should also work when we are not in a git directory (for example a build directory created with [wiki:Building/Using lndir]). -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9543> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 2

[GHC] #9253: "internal error: evacuate(static): strange closure type 0" when running in QEMU
by GHC 25 Apr '16

25 Apr '16

#9253: "internal error: evacuate(static): strange closure type 0" when running in QEMU ------------------------------------+------------------------------------- Reporter: ocharles | Owner: simonmar Type: bug | Status: new Priority: normal | Milestone: Component: Runtime System | Version: 7.8.2 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: Runtime crash Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | ------------------------------------+------------------------------------- We're trying to use NixOS to automatically run our acceptance tests in a VM on commit, in order to guarantee system stability. However, we're periodically seeing our web server component, compiled with GHC 7.8.2 is failing to start up. When it fails, the error message is always: {{{ internal error: evacuate(static): strange closure type 0 (GHC version 7.8.2 for x86_64_unknown_linux) Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug }}} Happy to do more debugging on this, but I'll need to be pointed in the right direction. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9253> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5

[GHC] #10725: Figure out how to support type synonym implementions of abstract data
by GHC 22 Apr '16

22 Apr '16

#10725: Figure out how to support type synonym implementions of abstract data -------------------------------------+------------------------------------- Reporter: ezyang | Owner: ezyang Type: task | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.2 (Type checker) | Keywords: backpack | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Revisions: | -------------------------------------+------------------------------------- A classic example we have used when describing Backpack is that you might want to abstract over is different types of strings: String, ByteString, Text, etc. Unfortunately, with every implementation of Backpack we have had so far, you can't actually do this! For example, say you use a type synonym to get an existing data type to the "right name"; you get this error: {{{ A.hs-boot:2:6: Type constructor `T' has conflicting definitions in the module and its hs-boot file Main module: type T = String Boot file: abstract(False) T No C type associated RecFlag NonRecursive FamilyInstance: none }}} How can we make this work? Could it be as simple as just relaxing the restriction here? This seems to work OK with shaping, so long as we refer to the type synonym (and not the original type in question!) If some type checker experts could weight in that would be quite useful. ---- Probably the biggest danger is interaction with machinery that works with type families. For example, this is accepted: {{{ -- A.hs-boot module A where data T -- B.hs {-# LANGUAGE TypeFamilies #-} module B where import {-# SOURCE #-} qualified A data T = T type family F a :: * type instance F A.T = Bool type instance F T = Int }}} This doesn't lead to unsoundness today, because A.T must always be defined in A.hs (and thus have a unique nominal identity). But it's dangerous for this to be accepted today with signatures, since T could refer to B.T. (The moral of the story is that after shaping, you have to recheck all pairs of families; you can't assume that because they were typechecked earlier that they're OK.) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10725> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 3

[GHC] #8524: GHC is inconsistent with the Haskell Report on which Unicode characters are allowed in string and character literals
by GHC 19 Apr '16

19 Apr '16

#8524: GHC is inconsistent with the Haskell Report on which Unicode characters are allowed in string and character literals ----------------------------+---------------------------------------------- Reporter: oerjan | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.6.3 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: GHC rejects valid program Unknown/Multiple | Test Case: Difficulty: Unknown | Blocking: Blocked By: | Related Tickets: | ----------------------------+---------------------------------------------- GHC is inconsistent with the Haskell Report on which Unicode characters are allowed in string and character literals. (And I don't like either option, why leave out any characters in strings unnecessarily?) Examples from ghci 7.6.3 (also tested in lambdabot on irc): {{{ Prelude> "" -- Unicode char \8203, Format class. <interactive>:10:2: lexical error in string/character literal at character '\8203' Prelude> " " -- Unicode char \8202, Space class. "\8202" Prelude> "t\ \est" -- Unicode char \8202 in a string gap. <interactive>:14:4: lexical error in string/character literal at character '\8202' }}} My reading of http://www.haskell.org/onlinereport/haskell2010/haskellch2.html (section 2.2 and 2.6): * The report BNF token "graphic", which can be used in literals, includes indirectly many Unicode classes, but uniWhite is not one of them. Thus the only Unicode whitespace allowed to represent itself in literals is ASCII space. * Unicode formatting characters are not mentioned in the BNF that I can see, so are not allowed in literals. * String gaps are made out of the report BNF token whitespace, which ''does'' include uniWhite. Who wants what: || ||= GHC =||= Report =||= Me =|| || Format in string || No || No || Yes || || Space/uniWhite in string || Yes || No || Yes || || Space/uniWhite in string gap || No || Yes || Dunno || In short, GHC's behavior is buggy and/or annoying in two opposite ways: * It leaves out some Unicode characters as allowable in strings and character literals, presumably because the report says so. * It allows some characters the report says it ''shouldn't'', and refuses some characters the report says it ''should''. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8524> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 9

[GHC] #11180: A program writing to a read-only stdout should not succeed
by GHC 19 Apr '16

19 Apr '16

#11180: A program writing to a read-only stdout should not succeed -------------------------------------+------------------------------------- Reporter: thomie | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.2 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: -------------------------------------+------------------------------------- In the shell, redirection of file descriptors can be read-only (<), write- only (>), or read-write (<>). A program trying to write to stdout, while stdout is redirected read-only, currently exits with exit code 0. {{{ $ cat hello.hs main = print "hello world" $ ghc hello.hs ... $ ./hello 1< /dev/null; echo Exit status: $? Exit status: 0 }}} This should however show some error message, and return some non-zero error message, because the write never succeeds. Note that this example is different from `./hello 1> /dev/null`, which rightfully does succeed. As suggested by this [http://roscidus.com/blog/blog/2013/06/09/choosing-a -python-replacement-for-0install/ blog post]: > "If you’re not sure why this is important, imagine the command is `dump-database > backup` and the filesystem is full." Supposedly OCaml gets this right. For reference, from my bash man page: > *Redirecting Input* > > Redirection of input causes the file whose name results from the expansion of word to be opened for reading on file descriptor n, or the standard input (file descriptor 0) if n is not specified. > > The general format for redirecting input is: > > [n]<word -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11180> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 5