Closed
Milestone
expired on Aug 1, 2018
8.6.1
Unstarted Issues (open and unassigned)
91
- package environment files and the GHC API
- Confusing parser error in 8.6
- Typed holes inferring a more polymorphic type
- Abstracting out pattern into a pattern synonym fails with scary error
- s390x builds flood with -Wunused-label warnings
- Add tests for SIMD loads and stores
- coerce fails for Coercible type families
- evolve / improve Native Gen Format in Format.hs (especially in context of post simd cleanup)
- Flag for printing absolute paths in diagnostics
- Users guide: broken external links
- Flags missing documentation.
- FloatFnInverses seems to show some weird rounding/precision issues.
- Allocations values for some compile tests are way too hight
- T9675 inexplicably regressed in allocations due to text submodule bump
- Nullary constraint in GHCi breaks `:t` command
- Simpliify tcTyConScopedTyVars
- Surprising failure combining QuantifiedConstraints with Coercible
- Undependable Dependencies
- Optimise wakeups for STM
- Speed up Data.Unique
- List comprehension seems to prevent some rewrite rules to fire
- Don't float out lets in between lambdsa
- Documentation of atomicModifyMutVar# does not show properly
- ref6 example from StaticPointers documentation doesn't type check
- PAP invariant of pointer tagging does not hold in profiling builds
- Stack overflow in withArgs leads to infinite memory-consuming loop
- Always promoting metavariables during type inference may be wrong
- Panic when abusing kind inference
- Hadrian puts its build tree in the wrong place
- C wrappers for Haskell foreign exports don't have finalizers (causes memory leak).
- security of package environment files
- `-XStrict -XNoStrict` is not neutral
- TypeInType: Type error conditioned on ordering of GADT and type family definitions
- COMPLETE pragmas make overlapping-patterns warnings behave oddly
- GHC.IO.Encoding not flushing partially converted input
- DuplicateRecordFields not honored within a data family?
- GHCi does not follow the XDG Base Directory Specification
- CApiFFI generates bad prototypes for pointers of `Foreign.C` types
- Cannot bind symbolic names in a rule
- ghci: dynamic linking fails on osx
- -XStrict doesn't prevent warnings from -Wunbanged-strict-patterns
- HPC: do notation marks () as non-covered
- Cannot install GHC through stack on NixOS
- GHC uses 300% CPU when calling into blocking C call
- T11627[ab] fail on some Darwin environments
- AArch64 validation fails with many invalid relocations
- Functions in HsUtils don't have the most general type
- Show instance for TypeReps is a bit broken
- Change in specialisation(?) behaviour since 8.0.2 causes 6x slowdown
- Lock .tix file
- Speed up the RTS hash table
- T12227 is failing on ghc-8.0
- T9405 fails on Windows
- Shutdown interacts badly with requestSync()
- InteractiveEval error handling gets a boot ModSummary instead of normal ModSummary
- TH_spliceE5_prof is failing with release candidate 8.0.1
- SourceNotes are not applied to all identifiers
- Cross-module SpecConstr
- Make the runtime reflection API for names, modules, locations more systematic
- The impossible happened: primRepToFFIType
- Binary size increase in nofib/grep between 7.6.1 and HEAD
- Some GHC.* module should export word size and heap object header size
- SingleEntry update flag for Stg bindings is not used
- Specialize min/max functions for GHC provided instances.
- Make bindings with multiple occurrences a join point instead of duplicating code during inlining.
- Expand comment in hPutStrLn
- Explore moving from pretty to prettyprinter
- Add some benchmarks to nofib from Andras Kovac's Eff benchmarks
- Implement -fexternal-interpreter via static linking
- More liberally kinded coercions for newtypes
- Type synonym unused binds no warning?
- GADT's displayed type is misleading
- Break up the stable pointer table
- Support promotion of pattern synonyms to kinds
- Add (or document if already exist) ability to derive custom typeclasses via source plugins
- ancient ghc release history on web page is incomplete
- Reduce type families in equations' RHS when testing equation compatibility
- A function `nat2Word# :: KnownNat n => Proxy# n -> Word#`
- Access the number of bits in the target machine's Int type at compile time
- Relaxing Levity-Polymorphic Binder Check for Lifted vs Unlifted pointers
- Provide a strict version of sum
- How to pass "-hide-all-packages" to the GHC API?
- Rewrite rules should be able to produce custom compiler errors
- Elide empty dictionaries
- Typeclass resolution errors quite puzzling
- Unboxed sums are not Typeable
- Missed constant folding oportunities
- Add some weird Kmettian tests to the test suite
- Allow documentary type signatures
- REMOVED pragma
- Deprecate exports
Ongoing Issues (open and assigned)
23
- Run sanity checker in more testsuite runs
- Unboxed tuples/sum error message on `a = show 5` in expression evaluation and interactive modes
- traceEvent tests failing in slow validate
- GHC boot script can't handle Git remote not named origin
- gcdExtInteger violates assertion
- Huge increase of compile time and memory use from 8.0.2 to 8.2.2 or 8.4.2
- Plugins don't work with -fexternal-interpreter
- tagToEnum# . dataToTag# not optimized away
- Simplify runghc command line options
- 20% regression in max_bytes_used for T1969
- linux/powepc : ghc-stage1 segfaults when building unregisterised
- Constraint vs *
- Document prim-ops
- explicit promotions of prefix data constructors can't be parsed naturally
- CUSK mysteries
- Asynchronous exception discarded after safe FFI call
- control what sort of entity a deprecated pragma applies to
- Refactor of Coercion
- Implement KPush in types
- Redesign the dynamic linking on ELF systems
- Support unconventionally named import libraries
- Statistical profiling support in the RTS
- Add more bit fiddling functions to 'integer-gmp'
Completed Issues (closed)
448
- Wrong Endian, ghc-pwd and ghc-cabal on ghc 8.0.1 for powerpc64le (IBM Power)
- topNormaliseType is wrong
- Constant folding optimises 1 into 3
- Non-deterministic segmentation fault in cryptohash-sha256 testsuite
- Plugin recompilation check fails when profiling is enabled
- Plugin recompilation check still panics
- GHC 8.6+ loops infinitely on an UndecidableInstances error message
- Merge llvm fix to ghc-8.6
- GHC 8.6.1 regression (buildKindCoercion)
- Parse error on input ‘*’
- Figure out why CircleCI isn't generating haddocks
- Cannot parse `type (*)` in fixity declaration
- Regression in Natural desugaring
- Unboxed Sums Crash
- Can't parse ''(*) in GHC HEAD
- WARNING in hptSomeThingsBelowUs when using a source plugin
- Apparent missing dependency in ghc-heap
- Tests are failing with segfaults under CircleCI
- GHC HEAD crashes on Windows 64 after the SRT overhaul (May 16, 2018 and newer)
- Slowdown in ghc compile times from GHC 8.0.2 to GHC 8.2.1 when doing Called arity analysis
- Segmentation fault of ghc-pkg.exe from 32-bit distribution of ghc-8.2.2 on Windows 7
- GHC HEAD regression: tcTyVarDetails
- High CPU when asynchronous exception and unblocking retry on TVar raced
- GHC HEAD typechecker regression
- ghc 8.4.2 on OS X: clang: warning: argument unused during compilation: '-nopie'
- print022 testcase fails on i386
- -split-objs generates excessively many files on Windows
- GHC HEAD regression: GADT constructors no longer quantify tyvars in topological order
- Improve performance of Simplify.simplCast
- CI with CircleCI
- 8.2.1 regression: heap corruption after safe foreign calls
- Core lint in BuildFlavour=perf-llvm
- Need -fno-print-explicit-runtime-reps to work on IfaceType, else RuntimeRep leaks
- Different saturations of the same polymorphic-kinded type constructor aren't seen as apart types
- GHC panic from pattern synonym, "Type-correct unfilled coercion hole"
- Marking Pred(S n) = n as injective
- Use String rather than [Char] where possible
- Dependent type synonym in recursive type family causes GHC to panic
- Core Lint error
- Errors about ambiguous untouchable variable when using constraint variable in RankN type
- ghc: panic! (the 'impossible' happened)
- Implication constraint priority breaks default class implementations
- Lost opportunity to eliminate seq
- panic! Simplifier ticks exhausted
- Higher-rank kinds
- Unused binding warning should not apply to newtype constructors
- Bug when using TimerManager/GHC.Event ?
- Got Unable to commit 16777216 bytes of memory error on Ubuntu
- No skolem info panic: Static pointers and holes
- RebindableSyntax warns `return` is not in scope when trying to call `pure`, but does not desugar between the two
- Profiling RTS flag `-po` does not work
- Unexpected compile error on type level lists with a single element
- Bug while writing a simple parser with Parsec
- --abi-hash with Backpack incorrectly loads modules from dependent packages
- QuantifiedConstraints: trouble with type family
- Compilation panic! (the 'impossible' happened)
- Testing
- Treating Constraint as Type when using (type C = Constraint)
- Phabricator shows "drafts" by default
- Unable to commit [negative number] bytes of memory
- TH improperly converts promoted data cons in ConT
- Core transformations generate bad indexCharOffAddr# call
- Plugin recompilation check panics with inline-java
- Typo in the documentation for Numeric.Natural.
- Inconsistent kind output for (->)
- Names of RULES aren't quoted in -ddump-splices
- Forced to enable TypeInType because of (i ~ i)
- Regression since 7.10.3, "variable not in scope" errors sometimes get buried under a sea of much less relevant errors.
- Panic when building Crucible
- TypeApplications error message doesn't parenthesize infix name
- Unicode 8.0 and later characters are invariably lexical errors
- GHC Panic on malformed newtype and StrictData
- Invalid warning Defined but not used
- -ddump-splices pretty-prints LambdaCase nonsensically
- GHCi internal error with runQ [| it |] and _
- `showEFloat` inconsistency introduced in base-4.12
- Deriving with QuantifiedConstraints is unable to penetrate type families
- Assertion failures in tests T7224, T9201, LevPolyBounded
- -ddump-splices truncates Integer literals to Int literals
- internal error: Unable to commit 1048576 bytes of memory. Deepseq
- if a variable that isn't in scope is used and main is not defined ghc freaks out
- No control over synopsis in Library Documentation
- "Ambiguos occurrence" error message uses strange qualification
- primops.txt.pp still has support for WORD_SIZE_IN_BITS < 32
- the_gc_thread variable from GC.c is not aligned to 64
- Validation doesn't play nicely on a shared computer
- Misleading error message on mising BangPatterns extension
- Bug in opt_trans_rule in OptCoercion
- Fix Git commit ID detection in worktrees
- Inconsistency w.r.t. coverage checking warnings for EmptyCase under unsatisfiable constraints
- compiling files in ghci on MacOS eventually results in malformed mach-o: load commands size (32800) > 32768
- SPECIALIZE one of two identical functions does not fire well
- Misleading error message for higher rank type
- Coercible and Existential types don't play nicely
- Oversaturated type family application panicks GHC (piResultTys2)
- Ungrammatical error message involving a pattern binding
- GHCi debugger doesn't see free variables when using ApplicativeDo
- Refactor (~) to reduce the superclass stack
- GHCi's :kind doesn't work with wildcards
- Poor error message when trying to import non-existent class methods
- "Instance head is not headed by a class" when `Constraint` replaced with `type C = Constraint`
- Fix typos in docs
- Spellcheck TcUnify file
- Spellcheck TcType file
- Fix a small typo in TcType.hs
- Typo in trac ticket number
- Weird GHCi bug
- Build failure on PowerPC 64-bit big endian
- GADT deriving Ord generates inaccessible code in a pattern with constructor.
- -staticlib is broken on GHC 8.4.3 on Linux under certain conditions
- Regression: Detection of unused imports is imprecise (since GHC 8.0.1)
- Unable to set testsuite verbosity to zero from command line
- TTG typo: XFieldOcc should be XCFieldOcc
- -Wincomplete-patterns gets confused when combining GADTs and pattern guards
- Infinite typechecker loop in GHC 8.6
- Testsuite: compile_timeout_multiplier is fragile
- GHCi leaks in DEBUG mode
- Typed hole panic on GHC 8.6 (tcTyVarDetails)
- Misleading error message when unboxed tuples used in sums without -XUnboxedTuples
- Argument-less infix declarations printed without parentheses in -ddump-splices
- Error message displays redundant equality constraint
- Functional dependencies can get GHC to print "UnkSkol"
- Another Cabal submodule bump
- False kind errors with implicitly bound kinds in GHC 8.6
- fixST is a bit wrong
- Core Lint error in GHC 8.6.1: From-type of Cast differs from type of enclosed expression
- Implicitly quantifying a kind variable causes GHC 8.6 to panic (coercionKind)
- :info prints kinds in closed type family equations without enabling -fprint-explicit-kinds
- -ddump-splices does not parenthesize visible type applications correctly
- -ddump-splices does not parenthesize rank-n contexts correctly
- mkGadtDecl does not set con_forall correctly
- Typed holes in Template Haskell splices produce bewildering error messages
- Core Lint error involving newtype family instances with wrappers
- Renamer plugins could run after each group has been renamed
- Internal error during typechecking of a hole in GHCi when there's shadowed identifiers
- Error message prints explicit kinds when it shouldn't
- Incorrect -ddump-deriv parenthesization for GND'd fmap implementation
- First attempt at starting GHCI failed
- API Annotations lost when parsing tyapp
- TTG for IPBind wrong extension name
- GHCi :print produces variables that cause a panic
- Unused "foralls" prevent types from being Coercible
- QuantifiedConstraints: panic "addTcEvBind NoEvBindsVar"
- isUnliftedType GHC panic on pattern with True :: Maybe
- "strange closure type" in T7919 with the threaded2 way
- Document how equality-bearing constructors are promoted in Core
- GHC 8.6.1 can't be bootstrapped with GHC 8.2.1
- CPP #includes may result in nonsensical SrcSpans
- Bogus type in typechecker error recovery
- Numerous validation failures when building GHC with LLVM
- Datatypes with CUSKs should quantify over unknown kinds
- Add QuantifiedConstraints to release notes
- Out-of-scope-variables error still can be deferred in ghci
- overflow1 breaks on i386
- Cabal submodule bump is needed
- Data family promotion is possible
- Ambiguity checks in QuantifiedConstraints
- -ddump-splices shenanigans: promoted tycons aren't ticked
- Typechecker sometimes doesn't preserve HsPar in original source.
- GHCi's :doc doesn't find the documentation for some class methods
- T13838 fails in ghci way
- New SRT scheme breaks -fvia-C
- GHCi pretty-prints (->)'s fixity poorly
- TypeError is reported as "redundant constraint" starting with GHC 8.2
- UndecidableInstances validity checking is wrong in the presence of QuantifiedConstraints
- unix tests fail under CircleCI's fedora environment
- Run typeCheckResultAction and renamedResultAction in TcM rather than Hsc
- GHC doesn't know that seq# produces something in WHNF
- `-fno-state-hack` produces incorrect results in nofib
- Serialising Names using the GHC API is not possible
- T14547 is broken
- HEAD doesn't build without sphinx-build
- Typo in documentation
- GHC HEAD internal error when FlexibleContexts isn't enabled
- Redefining (~) yields a baffling error message
- 32 bit Haddock runs out of memory compiling 32 bit GHC
- powerpc64le-unknown-linux is missing from llvm-targets
- ~# is always in scope with TypeOperators
- Strictness annotations bind more tightly than doc strings on non-record data declarations
- GHC 8.4 fails to build on Debian mips
- Internal error showing typed hole in GHCi
- Merge -XPolyKinds with -XTypeInType
- Catch cases where both branches of an if jump to the same block.
- HEAD: Validate failure when not TABLES_NEXT_TO_CODE
- ghc 8.4.2 panic in profiling build
- Levity Polymorphic type signatures in GHC.Prim
- got internal error when running a simple but computationally intensive loop
- Undecidable instances slip through
- Add HeapView functionality breaks validate (stage 0 ghc is 8.4.2)
- GHC HEAD panic (dischargeFmv)
- SRT offset optimisation breaks on MachO platforms
- Bad error message when `!` is used without BangPatterns extension
- GHC 8.2.2: internal error with +RTS -hb
- armv7l-unknown-linux-gnueabihf is missing from llvm-targets
- unregistersied GHC can't be built with ghc-HEAD: has no member named 'srt'
- T8089 failing with ghci, threaded1, threaded2, profthreaded
- Core Lint error in ill-typed GADT code
- Difference between 'libicuio' and 'icuio' in the 'extra-libraries' field
- Identical distinct type family fields miscompiled
- Documentation: ScopedTypeVariables feature not mentioned/misleadingly described
- Type inference regression between GHC 8.0.2 and 8.2.2
- Passing an IO value through several functions results in program hanging.
- GHC 8.4.2 installation crash with Stack
- EmptyCase produces incorrect SrcSpans
- enumFrom... aren't really documented.
- Hadrian-built GHC cannot locate `unlit`
- Speed up certain Foldable NonEmpty methods
- emitPrimOp: can't translate PrimOp
- Typeclass instance selection depends on the optimisation level
- make install fails in 8.5
- Program involving existentials and type class constraint gets rejected
- Printing non-ASCII characters to console on Windows
- GHC internal error when GADT return type mentions its own constructor name
- `Numeric.showEFloat` does not honor precision with 0 digits
- ghc 8.4.1 lifts errors out of if then else
- GHCi leaks the first modules loaded
- Exitification abstracts over shadowed variables
- Panic: collectNBinders
- Testsuite failures on Windows
- Update JMP_TBL targets during shortcutting for x86 codegen.
- GHC panic when compiling gi-pango-1.0.15
- $tooldir not expanded for NoFib
- Missing instance Applicative (K1 i c)
- pretty-show's literate haskell Setup.lhs failed on 32-bit windows
- unknown symbol `___divmoddi4' error with clock on 32-bit windows
- Testsuite should output JUnit XML for better CircleCI support
- Internal Error with Bibliography Profiling
- hT profiling option
- cannot install GHC on CentOS
- GHC HEAD regression: cannot instantiate higher-rank kind
- Unable to newtype derive `Prim` via DerivingStrategies
- When Typeable and unboxed sums collide, GHC panics
- T3001-2 fails on i386 Linux
- num009 is incredibly platform-sensitive
- outofmem testcase sometimes crashes on i386
- scc001 unexpected passes in prof way on CircleCI
- Lint types created by newFamInst
- ghci - improve error on hidden package
- DeriveAnyClass instances may skip TypeError constraints
- NUMA not detected on Win10 with AMD Ryzen Threadripper
- Outdated comments in RnSource.hs
- Bump unix's version number to 2.8.0.0 in HEAD
- rnIfaceBndr implementation in RnModIface looks incorrect
- DTrace enabled GHC does not work as a bootstrap compiler on FreeBSD
- Bizarre pretty-printing of inferred Coercible constraint in partial type signature
- Running 1 twice, followed by a typed hole, in GHCi causes internal error
- STG Linter dies when linting primitive
- Panic when using StaticPointers with typed holes
- ScopedTypeVariable and RankNTypes don't scope throughout function body
- GHCi panics when attempting to load archives
- [1,1..0] /= [1,1..2]
- Cabal Installation error in Windows 10 machine for Haskell
- Remove WrappedMonad from Control.Applicative
- Type deduction panic
- Stackage HEAD detected a suspicious change
- ghc-pkg list crashes on Windows when unicode character is in the path
- Bump base's version number to 4.12.0.0 in HEAD
- "Iface type variable out of scope" when compiling with -c
- Don't keep shadowed variables in ghci, both renamer and type checker
- GHC HEAD regression involving TYPEs in type families
- "Valid refinement suggestions" have the wrong types
- GADTs don't seem to unpack properly
- Clang doesn't support -Wsync-nand
- Location in GHCi debugger prompt printed twice when default prompt is used
- GHC.Err.errorWithoutStackTrace produces stack trace when profiling enabled
- Underconstrained typed holes are non-performant
- QuantifiedConstraints: introducing classes through equality constraints fails
- Panic when exporting duplicate record fields from separate modules
- Compiling error related to rts/Stats.c
- TH eats 50 GB memory when creating ADT with multiple constructors
- Bump required autoconf version
- Compiler panic (piResultTy)
- Calling isByteArrayPinned# on compact ByteArray
- ghc-HEAD broke cross-compilation on multiple tagets due to --via-asm switch
- The (->) type doesn't survive a TH quote-splice roundtrip
- TH breaks the scoping of quoted default method implementations when spliced
- QuantifiedConstraints don't kick in when used in TypeApplications
- Consistent labeling of redundant, qualified imports
- -ddump-splices pretty-printing oddities with case statements
- Documentation for isLiftedTypeKind is incorrect
- QuantifiedConstraints: Can't deduce `c' from `(a, b)' and `a |- b |- c'
- Can't splice a partially applied PromoteTupleT type from Template Haskell
- -O2 forces -optlo-O3 in a way that cannot be overridden at command line
- GHC 8.2 does not accept British {-# LANGUAGE GeneralisedNewtypeDeriving #-} spelling
- GHC 8.4.1 pretty-prints data family instances with redundant kind signatures using -ddump-splices
- GHC does not pass -dstg-lint
- Pretty Printing: GHC doesn't parenthesise (() :: Constraint)
- eqTypeRep does not inline
- accumArray is too lazy
- panic when using a function defined in terms of `error`
- MultiWayIf makes it easy to write partial programs that are not catched by -Wall
- Incorrect diagnostic for UNPACK/missing strictness
- DisambiguateRecordFields fails for PatternSynonyms
- GHC Panic with QuantifiedConstraints
- Hidden package hints no longer display
- GHCi doesn't load project.
- GHC 8.4.1-alpha allows the use of kind polymorphism without PolyKinds
- T11489 fails if run as root
- Runtime is keeping hold to too much memory
- ghc --interactive error: "gcc: error: libgmp.so: No such file or directory"
- GHCI doesn't recover from bad :add
- Compiling a function with a lot of alternatives bottlenecks on insertIntHeap
- Nofib: Ignore cr-lineendings in benchmarks
- Nofib: Strip called without .exe extension resulting in errors.
- Work out why Hadrian builds routinely fail
- refutable let bindings produce "irrefutable pattern failed" errors
- Wrong warning by -Wincomplete-patterns
- doCorePass has at least one missing case
- Ghc panics while building stage2 with -dstg-lint
- Refactor ConDecl
- Exponential runtime performance regression in GHC 8.2 + Data.Text.Lazy + Text.RE.TDFA
- Nofib is broken on Windows
- hp2ps and unlit are compiled by the stage0 compiler
- Tiered memory allocation restricts available memory
- Can't Hide Field When DuplicateRecordFields Is Enabled
- Certain do blocks cause TH to barf when ApplicativeDo is enabled
- Windows runtime linker failure with icuuc
- Need proper SCC analysis of type declarations, taking account of CUSKs
- Linker limit on OS X Sierra breaks builds for big projects
- LambdaCase should be suggested if it is used but not enabled
- Profiling breaks determinism
- `make fasttest` exits successfully even in presence of failures
- `make binary-dist` is broken
- bad pretty-printing of types with promoted data types
- Head does not build on Windows with the default make config.
- Assertion triggered by STM invariant.
- `:kind` suppresses visible dependent quantifiers by default in GHCi 8.2.1
