Closed
Milestone
expired on Jul 26, 2017
8.2.1
The 8.2.1 release will be the second major release in the GHC 8 series.
See also [[Status/GHC-8.2.1]] for more details.
Migration Guide: [[Migration/8.2]]
Unstarted Issues (open and unassigned)
19
- Killing a thread will block if there is another process reading from a handle
- very slow linking of executables with ld.bfd < 2.27
- Concurrent modifications of package.cache are not safe
- Partially applied constructors with unpacked fields simplified badly
- GHC 8.0.1 build segmentation fault in haddock
- Fix GHC on AArch64/Arm64
- LLVM/optimized code for sqrt incorrect for negative values
- Native Code Generator for 8.0.1 is not as optimized as 7.6.3...
- installed include/HsVersions.h wants to #include "../includes/ghcautoconf.h"
- GHCi is setting stdin/stdout to NoBuffering in runghc when DYNAMIC_GHC_PROGRAMS=YES
- Exception: ToDo: hGetBuf - when using custom handle infrastructure
- Generalize the type of Data.List.{deleteBy, deleteFirstsBy}
- Invalid Eq/Ord instances in Data.Version
- Unbundle Cabal from GHC
- Add proper support for weak symbols to the runtime linker
- The GHC API doesn't provide a good hscTarget option for tooling
- Feature request: Add index/read/write primops with byte offset for ByteArray#
- Make the rewrite rule system more powerful
- Add richer data model address arithmetic: AddrDiff and AddrInt (ie d Int_ptr_diff and Int_ptr_size)
Ongoing Issues (open and assigned)
7
- Data constructors should not have RULES
- GHCi leaky abstraction: error message mentions `ghciStepIO`
- AArch64 : ghc-stage2 segfaults compiling libraries/parallel
- Threaded RTS performing badly on recent OS X (10.8?)
- "make install-strip" should work
- GHC should delete all temporary files it creates in /tmp
- AArch64: get GHC Calling convention working
Completed Issues (closed)
546
- Don't use gtar to produce FreeBSD binary distribution
- Segmentation fault inside threadPaused
- Optimizations create run time seg faults
- Bump Cabal submodule
- MonoLocalBinds/RankNTypes type inference regression in GHC 8.2
- Runtime crash with <<loop>> after concurrent stressing of STM computations
- xmobar crashes with segmentation faults?
- Internal libraries regression in GHC 8.2 (ghc-pkg handling)
- Bump time submodule
- join points produce bad code for stream fusion
- Regression: can't export constructor when conflicting, qualified constructor is also in scope
- Nondeterminism in ‘pure’ function w/ parallel evaluation & memo combinators
- regression: `ANN` pragmas no longer ignored in TH-less GHC 8.2.1
- Something is fishy in Windows hLock implementation
- Update Cabal submodule to whatever the final Cabal 2.0 release is
- ala from Control.Lens.Wrapped panics
- Incorrect Record Pattern Synonym example in users guide
- Windows binary distributions have absolute paths from the build environment in the `settings` file
- Make it easier to use the gold linker
- Program which terminates instantly in GHC 8.0.2 runs for minutes with 8.2.1
- hWaitForInput with timeout causes program to abort
- GHCi preprocesses every source file on :reload
- GHC HEAD regression: Prelude.!!: index too large
- Panic with unsafeCoerce and -dcore-lint
- Fix 32 bit Windows build of GHC 8.2
- Segmentation faults in profiled way
- Optimizer produces Core with an infinite <<loop>>
- GHC HEAD panic: collectNBinders
- GCC driver fails on Windows 10 15019+
- GHC HEAD regression: Template variable unbound in rewrite rule
- Windows build thinks sqrt (-1) == -1
- ghc-pkg "cache is out of date" message breaks testsuite
- Package database locking patch broke ghc-pkg with non-existent database
- COMPLETE sets aren't read from external packages
- Make GHC handle orphan COMPLETE sets of conlikes better
- Core top-level bindings no longer deduplicated
- New versions of time and Cabal are causing a Core Lint error on Windows
- Typeable regression in GHC HEAD
- perf-llvm build fails with "Too many sections: 123418 (>= 65280)"
- cross-endian GHC generates invalid .hi files
- aws package fails to build with master
- Panic "StgCmmEnv: variable not found" with ApplicativeDo and ExistentialQuantification
- Perplexing levity polymorphism issue
- scrutinee constant folding produces wrong answer when remapped values are out of bounds
- Core Lint error on git HEAD
- maintainer-clean should clean build.mk
- High memory usage during compilation
- ghc master (8.1.20161206) panics with assertion failure with devel2 flavor and -O
- Panic Simplifier ticks exhausted since ghc 8
- master branch broken with alex 3.2
- Profiled build broken
- regression in type inference with respect to type families
- Type checker rejects valid program
- Type inference regression with RankNTypes (GHC 8.1)
- Compact regions support needs some discussion in the release notes
- Non-deterministic ghc-iserv terminated error
- Get rid of extra_files.py
- Recursive use of type-in-type results in infinite loop
- Disabling idle GC leads to freeze
- runST with lazy blackholing breaks referential transparency
- Kind of (->) type constructor is overly constrained
- Pattern matching against sets of strings sharing a prefix blows up pattern checker
- Combine exprIsTrivial and cpe_ExprIsTrivial
- 7.6.2 Segmentation Fault/Bus Error in large exponentation
- GHC panic from emacs haskell mode
- sortOn with tuple of >= 9 Maybe's triggers ghc: panic! with "Simplifier ticks exhausted"
- panic! (the 'impossible' happened)
- ghc panic: -fexternal-interpreter + .o + QuasiQuotes?
- Panic when using a TH splice in a do bind pattern
- ExnStr doesn't propagate "outwards"
- Official FreeBSD 10.3 distribution should link against libutil.so.9, not libutil.so.8
- GHC incorrectly suggests a data constructor is in scope in 8.2
- STG linter's type equality can loop
- @progbits is used incorrectly throughout GHC
- AC_CHECK_FUNC yields incorrect result for pthread functions on NixOS causing build to fail
- FP_CONVERT_CPU fails to convert CPU when invoked from bindist
- GHC 8.2 regression: "Can't find interface-file declaration" for promoted data family instance
- Improve [-Wmissing-home-modules] error message
- isByteArrayPinned# should consider BF_LARGE
- Template Haskell pretty-printer doesn't parenthesize infix datatype names in data declarations
- Inexplicable error message when using out-of-scope type variable in pattern type signature
- panic initTc: unsolved constraints
- Strange interaction between higher-rank kinds and type synonyms
- ApplicativeDo desugaring is lazier than standard desugaring
- GHC panic in 8.2 only: typeIsTypeable(Coercion)
- -g doesn't work with -pgma=clang++
- RTS Stats are recorded without -T
- Incorrect function signature causes the impossible to happen
- No parameter-validation in Control.Concurrent.setNumCapabilities
- "What AbsBinds means" is a pretty confusing doc comment
- GHC not using injectivity?
- deriveConstants: no objdump program given (OpenBSD)
- Gold linker fails
- GHC 8.2 nondeterministic with foreign imports
- Panic while forcing the thunk for TyThing IsFile (regression)
- Document allowed syntax in WARNING and DEPRECATED pragmas
- Infinite loop in compiler without undecidableXXX
- Bullish use of Template Haskell's newName causes GHC internal error
- TypeInType is not documented in the users' guide flag reference
- Deriving can't find an instance that holds, standalone deriving works
- Typo in user guide suggests that there's an -O* option
- `-Wmissing-home-modules` doesn't properly recognize filepath-qualified modules
- Panic! (the "impossible" happened) bug in GHC 8.2.1 rc2
- Failure of improvement for type-family dependencies
- Highly nested UNPACKed data causes panic
- Incorrect parenthesization of type in error message
- Broken Link in GHC Docs for Custom compile-time errors
- Ill formatted warning about tabulators
- Type inference fails in connection with TypeFamilies, 'let', type equalities ...
- Assertion failure on HEAD: "optCoercion changed types!"
- Invalid redundant pattern matches with -XTypeFamilyDependencies
- RebindableSyntax causes type errors when 'fail' is not defined, even if not used.
- GHCi 8.2 simply ignores TH splice using datatype with a forall'd kind signature
- Frontend plugin arguments are reversed
- GHC internal error: ‘Y’ is not in scope during type checking, but it passed the renamer
- Reified data family instances type variables not related to value constructor fields
- Segfault due to levity polymorphism of mkWeak#
- Can't resolve levity polymorphic superclass
- Typechecker behavior w.r.t. BangPatterns and nested foralls has changed in 8.2
- "*** Exception: expectJust showModule" in ghci with hs-boot
- Name is reported as ambiguous and not in scope
- Typechecker regression when combining PolyKinds and MonoLocalBinds
- -ddump-splices doesn't parenthesize type/data families correctly in 8.2.1
- Weird kind inference problems with closed type families
- ghc-8.2.0.20170404/testsuite/timeout failed to build
- Unqualified export of constructors doesn't work anymore
- ghc has a stack space leak when prints warnings
- -Wsimplifiable-class-constraints behaves differently with OVERLAPPING and OVERLAPPABLE
- GHC does not preserve order of forall'd vars with TypeApplications
- Unexpected failure of `annrun01` on 32-bit Windows
- Perplexing type error with unboxed tuples
- Clarify Some Restrictions on Compact Regions
- ghc-stage1: panic! (the 'impossible' happened) creating cross compiler for PPC
- ASSERT failed in compiler/typecheck/TcErrors.hs
- UNIQUE_BITS is not crosscompiler-friendly
- openTempFile fails when the template parameter contain a path separator
- GHC panic with deferred custom type errors
- Core Lint issues during slowtest
- Variables in patterns made with QuasiQuotes sometimes don't get bound
- `gcc.exe' failed in phase `Assembler'. (Exit code: -1073741502)
- Pattern synonyms bind variables out of scope
- GHC stubbornly produces dead code for empty case with type family
- Ghci panics with type applications to unknown functions
- unsafeInterleaveST is too unsafe
- Deferred type error sneaks in with -fno-defer-type-errors enabled
- Disable diagnostic coloring by default
- Type inference failure in bidirectional pattern synonym and GADT pattern match
- Segfaults on levity-polymorphic type class
- hs_try_putmvar003 is timing out / segfaulting
- Semantically-equivalent expressions evaluating at different values with -O1
- shebang + literate haskell causes line numbering skew
- Derive instances for classes with associated types
- GHCi gives conflicting information about visible type application
- realgcc.exe do not run GHCi in Windows in a particular case
- Associated type family instance validity checking is too conservative
- 3x slowdown on GHC HEAD with file containing lots of overloaded string literals
- Caret diagnostics interact badly with hsc2hs
- Segfault when profiling on OS X
- ghci fails to start when -XRebindableSyntax is passed
- Join ceilings incorrectly getting placed outside value lambdas by SetLevels
- "opt_univ fell into a hole" panic from rewrite rule
- GHC fails to specialize a pair of polymorphic INLINABLE functions
- instance selection too eager
- addCStub doesn't allow control over compiler flags or source file file
- Core string literal patch regresses compiler performance considerably
- Levity polymorphism-related GHC panic: expectJust zonkTcTyVarToVar
- GHC doesn't think a type is of kind *, despite having evidence for it
- Non-abstract types also have skolem nature
- forkIO has inconsistent behavior under optimization
- Binary distributions produced from cross-compiled stage2 are broken
- Unfortunate compiler loop when creating type loop (with UndecidableInstances)
- cabal08 broken
- GHC base directory assumptions
- Resident set size exceeds +RTS -M limit with large nurseries
- The runtime parses arguments past -- under windows but passes them on as arguments on linux
- Bug in GHC.Stack.callStack when used with sections
- test 'debug' is timing out on OS X
- DeriveAnyClass regression involving a rigid type variable
- Constraint synonym instances
- GHC panic with (->) generalization branch while compiling lens
- panic on unboxed string literal in pattern
- Backpack: matching newtype selectors doesn't work
- hPutBuf issues unnecessary empty write syscalls for large writes
- Default FD buffer size is not a power of 2
- Surprising inlining failure
- OccurAnal fails to rediscover join points
- Performance regressions in testsuite from join points
- <$ is bad in derived functor instances
- Orphan instances in Backpack signatures don't work
- Error in GHC.Generics documentation
- Core Lint error running testsuite with DEBUG way
- Levity polymorphism panic in GHCi
- incorrect links and file names in GHC user's guide
- Document AMP as a deviation from Haskell 2010
- COMPLETE pragma causes compilation to hang forever under certain scenarios
- Strange debug output printed: isPredTy Const Int (,) Bool Char
- Template Haskell needlessly pretty-prints empty instance contexts
- Fix mismatch between unsafeDupablePerformIO and note
- panic on negative literal in case on Word
- Make type import/export API Annotation friendly
- CSE missing an easy case
- NOINLINE and worker/wrapper
- Don't check for Perl in ./configure when not splitting objects
- Foreign.C.Types does not have a CBool type
- Typechecker "panic! the 'impossible' happened"
- Compilation fails with a panic: get_op runContT
- atomicModifyMutVar# has incorrect type
- Regression: TH splice requires TypeInType when it shouldn't
- Can't compile HEAD with ghc-7.10.1
- Runtime linker errors with CSFML on Windows
- Constraint solving failure with Coercible + type family + newtype
- GHCi segfaults with the flag -llibopenal, but not with -lopenal
- Fix typo in comment
- GHC panics can cause text-formatting changes from colorized error messages to linger
- BooleanFormula implies is wrong
- GHC should not allow modules to define instances of abstract type classes
- Abstract closed type families don't work with Backpack
- Backpack doesn't check for fixity consistency
- Deriving Foldable causes GHC to take a long time (GHC 8.0 ONLY)
- Adding INLINABLE pragma causes GHC panic
- Make `-dead_strip` the default on Darwin
- GHC 7.10->8.0 regression: GHC code-generates duplicate _closures
- GHC enters a loop when partial type signatures and advanced type level code mix
- The let/app invariant, evaluated-ness, and reallyUnsafePtrEquality#
- Type family reduction irregularity (change from 7.10.3 to 8.0.1)
- TH-spliced pattern synonym declaration fails to typecheck
- Timeout no longer kills processes on timeout
- libraries/base/System/Posix/Types.hs: tries to define Floating instance for opaque type
- Language.Haskell.TH doesn't reexport everything from Language.Haskell.TH.Lib
- GHCi incorrectly reports the kind of unboxed tuples spliced in from Template Haskell
- User guide section on AllowAmbiguousTypes should mention TypeApplications
- Template Haskell: unboxedTupleTypeName doesn't handle unboxed 0- and 1-tuples
- No levity-polymorphic arguments
- GHC panics with forall and constraint
- New haddock seems to be segfaulting
- GHC doesn't warn about missing implementations for class methods beginning with an underscore
- Properly detect if GHCi is running in MinTTY
- In a record-update construct:ghc-stage2: panic! (the 'impossible' happened)
- Something has broken T12903
- Confusing error (module is not loaded) when more hsigs provided than -instantiated-with
- unexpected uncaught segmentation error in arbitrary precision Integer calculations with modestly large numbers
- Cannot load hmatrix with ghci -fexternal-interpreter: missing symbol ___ieee_divdc3
- Backpack signature matching doesn't pick up orphan instances
- Type inference regression in GHC HEAD
- Colorized error messages result in hard-to-read .dump-* files
- GHC panic: Can't serialise IfaceTcTyVar
- initTc: unsolved constraints
- Make DefaultSignatures more particular about their types on the RHS of a context
- Python 3.2 doesn't support Unicode literal syntax
- Core lint failure with pattern synonym and levity polymorphism
- SIGINT is not handled in forked process
- Levity polymorphic expressions mustn't be floated out
- ghc panics at random places with -jX
- "too many iterations" causes constraint solving issue.
- (xs -> xs) wrongly suggests TypeApplications
- Internal error while typechecking invalid program
- GHC internal error while building darcsden
- POSIX osDecommitMemory has nasty compatibility problems
- Unpromoted tuples in TH in correctly accepted by the type checker
- Regression: GHC doesn't show filepaths by default anymore
- Panic with unboxed integers and ghci
- On Windows, runtime linker can't find function defined in GHC's RTS
- GHC-internal triplets passed to sub-project configure scripts
- mmap functions not properly guarded by RTS_LINKER_USE_MMAP in Linker.c
- On iOS, sys_icache_invalidate called with an end pointer instead of a size
- Pattern synonym signatures are not validity-checked
- Should GND infer an instance context when deriving method-free classes?
- GHC tells me to use RankNTypes when it's already enabled
- curl dependency in conjunction with Template Haskell causes "unknown symbol" errors
- prototype mismatch with EFF_ from unregisterisered GHC when building ieee754
- hs_try_putmvar003 sometimes fails
- Itimer.c doesn't compile on iOS non-threaded RTS
- hschooks.c #include path is incorrect
- GHC panic, `tcTyVarDetails` is missing a case
- Adding an explicit export list halves compilation time.
- --numa doesn't work even if libnuma is available
- Orphan hashes are not being propagated correctly
- T7037 is broken on Windows
- Bytes allocated statistic for T10858 differs between Windows and Linux
- GHC Internal error, unboxed sums
- GHC panic
- Improve output of pattern synonym info
- bkpcabal01 testcase fails with spurious output on stderr on Darwin
- Representation polymorphism validity check is too strict
- Program that fails Core Lint terribly
- Testsuite driver fails on Windows
- Type family reification misses a kind signature
- Overloaded method causes insufficient specialization
- DeriveAnyClass fails to derive some classes
- Support US spelling of `specialise` throughout
- Consistent use of sigs vs signatures in warnings
- Testsuite exhibits large amount of framework failures
- Inconsistent unicode display for kinds
- Panic on ":t datatypeName"
- Compilation time/space regression in GHC 8.0/8.1 (search in type-level lists and -O)
- Template Haskell boxes singleton unboxed tuples when splicing them
- Template Haskell regression: GHC erroneously thinks a type variable is also a kind
- UnboxedTuple makes overloaded labels fail to parse
- DEFAULT_TMPDIR is documented, but doesn't exist
- Conflicting definitions error does not print explicit quantifiers when necessary
- DeriveDataTypeable - deriving instance Data (Mu (Const ()))
- Pattern synonym parse error should recommend enabling extension
- With -O1 and above causes ghc to use all available memory before being killed by OOM killer
- Panic with DeriveAnyClass
- API annotations for unboxed sums needs reworking
- Fancy BinIface encoding for tuples is broken for constraint tuples
- Unboxed tuples have no type representations
- Template Haskell boxes unboxed tuple types when reifying them
- Panic when specializing in another module
- Type family not reduced
- WARN pragma gives warning `warning: [-Wdeprecations]'
- Demand Analyzer: Cunnig plan not adhered to with aborting fixpoint interation
- Increasing maximum constraint tuple size significantly blows up compiler allocations
- base/tests/topHandler03.run failures
- Dependent type family does not reduce
- 'deriving Eq' on recursive datatype makes ghc eat a lot of CPU and RAM
- Narrow the scope of record wildcards notation slightly
- ghc ../Test (without file extension) should work
- Windows linker stack commit setting causing issues
- Multiple pattern type signatures accepted
- Surprising error message when deriving Functor for a GADT
- GHC panic when using DeriveAnyClass with functor-like class and datatype with a type variable in a contravariant position
- SIGABRT on right-shift operation against long negative integer
- Type families, `TypeError`, and `-fdefer-type-errors` cause "opt_univ fell into a hole"
- Typed Template Haskell missing utilities provided by untyped variant
- ghc --show-options lists invalid flags
- ghc --help suggests -auto-all instead of -fprof-auto
- ghc-8.0.1-rc4: tyConRoles sees a TcTyCon
- DYNAMIC_GHC_PROGRAMS=NO: T8761 is failing (Make pattern synonyms work with Template Haskell)
- Typechecker panic instead of proper error
- Shutdown interacts badly with requestSync()
- Shutdown interacts badly with requestSync()
- should runRW# (of GHC.Magic) be reexported by GHC.Exts?
- Improve error message about closed variables
- Core lint error on record syntax update/pattern synonym
- Can't use IntPtr or WordPtr in a foreign import
- Surprising behavior with higher-rank quantification of kind variables
- USE_PTHREAD_FOR_ITIMER causes unnecessary wake-ups
- Associated pattern synonyms not included in haddock
- GHC forgets constraints
- Allow reify to yield types in the current declaration group
- panic "hscCmmFile: no_mod" with `-v2 and above
- `Strict` causes core lint error
- Simplifier: Inlining trivial let can lose sharing
- GHC panics ('unload: no interpreter") during T10052 test
- No TypeRep for unboxed tuples
- rts/posix/Itimer.c: ignoring return value of ‘read’
- Incorrectly suggests RankNTypes for ill-formed type "forall a. Eq a. Int"
- Alllow static pointers to local closed definitions
- Panic because of "updateRole"
- reify returns Dec that use ConT instead of PromotedT
- Cache coercion kinds and roles
- Self-kinded type variable accepted
- make -j24 test
- Regression in nofib/shootout/k-nucleotide
- Restore code to handle '-fmax-worker-args' flag
- Accessing shadowed definitions in GHCi
- GHC fails to compile if source is in /ghc
- ghc: panic! occured
- Using a dummy typechecker plugin causes an ambiguity check error
- Incorrect error message in StandaloneDeriving: "The data constructors of <typeclass> are not all in scope"
- `GHC.Base.eqString`'s documentation is messed up
- Building nofib/fibon returns permission denied
- testsuite should ignore config files
- Eliminate check_lifted check in TcValidity
- Associated types at wrong type in instance
- 8.0 rc1 panics in applyTypeToArgs
- Provide a way to make fast in-calls into Haskell
- DWARF call frame information incorrect in the presence of unsafe foreign calls
- Unwind information is incorrect in region surrounding a safe foreign call
- Unwind information incorrect between Sp adjustment and end of block
- CmmSwitchTest is broken on 32-bit platforms
- PowerPC NCG: Assign all STG float and double regs to PowerPC registers
- Overloaded state-monadic function is not specialised
- RebindableSyntax with pattern guards needs `fail`
- Testsuite framework failures are too easy to ignore and too hard to find
- New exhaustiveness checker breaks ghcirun004
- -O0 -g slows GHC down on list literals (compared to -O0 without -g)
- Type families + hs-boot files = panic (type family consistency check too early)
- GHC on Windows could not use more than 64 logical processors
- lookupTypeName does not support type operators not starting with :
- Split SRT tables on Mac OS
- ghc 7.10.2 testsuite: "DoParamM(normal)" test fails
- GHC Panic on import with 'OPTIONS_GHC -fobject-code -O'
- Periodic alarm signals can cause a retry loop to get stuck
- maximumBy has a space leak
- No non-exhaustive pattern match warning given for empty case analysis
- Handling of ASCII encodings introduced in D898 breaks Unicode terminal detection
- DeriveAnyClass and GND don't work well together
- Testsuite driver does not run tests in parallel on Windows
- AArch64 : divbyzero test fails
- Plugins/dynamic loading subtly broken (it seems)
- Literal Pattern match loses order
- panic in new integer switch logic with "out-of-range" literals
- ghc: panic! (the 'impossible' happened)
- Regression on MacOS platform, error in ghci calling main after loading compiled code: "Too late for parseStaticFlags..."
- Fix misattribution of Cost Centre profiles to lintAnnots
- DeriveAnyClass support for higher-kinded classes + some more comments
- Error when reifying type constructor
- readRawBufferPtr and writeRawBufferPtr allocate memory
- compile-time performance regression (probably due to Generics)
- String literals are wasting space
- Signed/unsigned integer difference between compiled and interpreted code
- LLVM backend TBAA is too aggressive
- CPP does not process test case enum01.hs correctly
