2.1. Version 9.12.1

The significant changes to the various parts of the compiler are listed in the following sections. See the migration guide on the GHC Wiki for specific guidance on migrating programs to this release.

2.1.1. Language

  • New language extension: OrPatterns implements GHC Proposal #522).

  • GHC Proposal #569 “Multiline string literals” has been implemented. The following code is now accepted by GHC:

    {-# LANGUAGE MultilineStrings #-}

x :: String
x =
  """
  This is a
  multiline

      string

  literal
  """

    This feature is guarded behind MultilineStrings.

  • The ordering of variables used for visible type application has been changed in two cases. It is supposed to be left-to-right, but due to an oversight, it was wrong:

    • in an infix application f :: a `op` b, it is now forall a op b. rather than forall op a b.

    • in a linear type f :: a %m -> b, it is now forall a m b. rather than forall a b m..

    This change is backwards-incompatible, although in practice we don’t expect it to cause significant disruption.

  • The built-in HasField class, used by OverloadedRecordDot, now supports representation polymorphism (implementing part of GHC Proposal #583). This means that code using UnliftedDatatypes or UnliftedNewtypes can now use OverloadedRecordDot.

  • Unboxed Float#/Double# literals now support the HexFloatLiterals extension (#22155).

  • UnliftedFFITypes: GHC will now accept FFI types like: (# #) -> T where (# #) is used as the one and only function argument.

  • The venerable default declarations have been generalized. They can now name a class other than Num and the class defaults can be exported. The functionality is guarded by the new NamedDefaults language extension. See the GHC proposal for details.

  • GHC now takes COMPLETE pragmas into consideration when deciding whether pattern matches in do notation are fallible.

  • As part of GHC Proposal #281 GHC now accepts type syntax in expressions, namely function type arrow a -> b, constraint arrow a => b, and forall telescopes:

    g = f (forall a. Show a => a -> String)
  where f :: forall a -> ...

  • In accordance with GHC Proposal #425, GHC now permits wildcard binders in type declarations:

    type Const a b = a   -- before: the `b` had to be named even if unused on the RHS
type Const a _ = a   -- now: the compiler accepts a wildcard binder `_`

2.1.2. Compiler

  • Constructor PluginProv of type UnivCoProvenance, relevant for typing plugins, gets an extra DCoVarSet argument. The argument is intended to contain the in-scope coercion variables that the the proof represented by the coercion makes use of. See Note [The importance of tracking UnivCo dependencies] in GHC.Core.TyCo.Rep, Constraint solving with plugins and the migration guide.

  • The flag -fprof-late will no longer prevent top level constructors from being statically allocated.

    It used to be the case that we would add a cost centre for bindings like foo = Just bar. This turned the binding into a CAF that would allocate the constructor on first evaluation.

    However without the cost centre foo can be allocated at compile time. This reduces code-bloat and reduces overhead for short-running applications.

    The tradeoff is that calling whoCreated on top level value definitions like foo will be less informative.

  • A new flag -fexpose-overloaded-unfoldings has been added providing a lightweight alternative to -fexpose-all-unfoldings.

  • -Wderiving-typeable has been added to -Wall.

  • SIMD support has been added to the X86 native code generator. For the time being, only 128 bit wide vectors are supported, with most floating-point operations implemented, together with a few integer vector operations. Other operations still require the LLVM backend. Contributors welcome!

  • i386 Windows support is now completely removed amid massive cleanup of legacy code to pave way for Arm64 Windows support (#24883). Rest assured, this does not impact existing support for x86_64 Windows or i386 Linux. For end users, the stdcall C calling convention is now fully deprecated and GHC will unconditionally produce a warning and treat it as ccall. All C import/export declarations on Windows should now use ccall.

  • 32-bit macOS/iOS support has also been completely removed (#24921). This does not affect existing support of apple systems on x86_64/aarch64.

  • The flag -fignore-asserts will now also enable the CPP macro __GLASGOW_HASKELL_ASSERTS_IGNORED__ (#24967). This enables people to write their own custom assertion functions. See Assertions.

  • The flag -fkeep-auto-rules that forces GHC to keep auto generated specialization rules was added. It was actually added ghc-9.10.1 already but mistakenly not mentioned in the 9.10.1 changelog.

  • Fixed a bug that caused GHC to panic when using the aarch64 ncg and -fregs-graph on certain programs. (#24941)

  • A new warning -Wview-pattern-signatures is introduced to notify users about future changes in parsing of view patterns in combination with pattern signatures

  • GHC now includes a new experimental flag, -fobject-determinism, which enables deterministic object code generation, with a minor trade-off in compilation speed (typically a 1-2% regression). Note, however, that this flag doesn’t yet ensure determinism across all compilation configurations; we anticipate these cases will be addressed in future updates (#12935).

2.1.3. GHCi

  • Fix a bug where GHCi would not start alongside a local file called Prelude.hs or Prelude.lhs (#10920).

2.1.4. Runtime system

  • Reduce fragmentation incurred by the nonmoving GC’s segment allocator. In one application this reduced resident set size by 26%. See #24150.

  • Memory return logic now uses live bytes rather than live blocks to measure the size of the heap. This primarily affects the non-moving GC, which should now be more willing to return memory to the OS. Users who have fine-tuned the -F ⟨factor⟩, -Fd ⟨factor⟩, or -O ⟨size⟩ flags, and use the non-moving GC, should see if adjustments are needed in light of this change.

  • The new runtime flag --read-tix-file=<yes|no> allows to modify whether a preexisting .tix file is read in at the beginning of a program run. The default is currently --read-tix-file=yes but will change to --read-tix-file=no in a future version of GHC. For this reason, a warning is emitted if a .tix file is read in implicitly. You can silence this warning by explicitly passing --read-tix-file=yes. Details can be found in GHC proposal 612.

2.1.5. Cmm

  • The bits128, bits256 and bits512 types have been removed, in favour of vec128, vec256 and vec512.

  • The [*] jump annotation (“all registers live”) has been removed, in favour of more specific annotations GP_ARG_REGS (all general-purpose registers live), SCALAR_ARG_REGS (all scalar registers live), and V16_ARG_REGS, V32_ARG_REGS and V64_ARG_REGS (meaning: all scalar registers plus all vector registers up to the given vector width in bytes).

2.1.6. base library

  • Propagate HasCallStack from errorCallWithCallStackException to exception backtraces, fixing a bug in the implementation of CLC proposal 164 <https://github.com/haskell/core-libraries-committee/issues/164>.

  • Add exception type metadata to SomeException’s displayException and “Exception:” header to the default handler (i.e. GHC.Conc.Sync.uncaughtExceptionHandler):

    https://github.com/haskell/core-libraries-committee/issues/231 https://github.com/haskell/core-libraries-committee/issues/261

  • The deprecation process of GHC.Pack <https://gitlab.haskell.org/ghc/ghc/-/issues/21461> has come its term. The module has now been removed from base.

  • Annotate re-thrown exceptions with the backtrace as per CLC proposal #202 <https://github.com/haskell/core-libraries-committee/issues/202>

  • Introduced catchNoPropagate, rethrowIO and tryWithContext as part of CLC proposal #202 <https://github.com/haskell/core-libraries-committee/issues/202> to facilitate rethrowing exceptions without adding a WhileHandling context – if rethrowing e, you don’t want to add WhileHandling e to the context since it will be redundant. These functions are mostly useful for libraries that define exception-handling combinators like catch and onException, such as base, or the exceptions package.

2.1.7. ghc-prim library

  • Usage of deprecated primops is now correctly reported (#19629).

  • New primops isMutableByteArrayWeaklyPinned# and isByteArrayWeaklyPinned# to allow users to avoid copying large arrays safely when dealing with ffi. See the users guide for more details on the different kinds of pinned arrays in 9.12.

    This need for this distinction originally surfaced in https://gitlab.haskell.org/ghc/ghc/-/issues/22255

  • New fused multiply-add instructions for vectors of floating-point values, such as fmaddFloatX4# :: FloatX4# -> FloatX4# -> FloatX4# -> FloatX4# and fnmsubDoubleX2# :: DoubleX2# -> DoubleX2# -> DoubleX2# -> DoubleX2#. These follow the same semantics as fmadd/fmsub/fnmadd/fnmsub, operating in parallel on vectors of floating-point values.

  • New vector shuffle instructions, such as shuffleFloatX4# :: FloatX4# -> FloatX4# -> (# Int#, Int#, Int#, Int# #) -> FloatX4#. These instructions take two input vectors and a collection of indices (which must be compile-time literal integers), and constructs a result vector by extracting out the values at those indices. For instance, shuffleFloatX4# on input vectors with components (# 0.1#, 11.1#, 22.1#, 33.1# #) and (# 44.1#, 55.1#, 66.1#, 77.1# #), and indices (# 4#, 3#, 6#, 1# #), will return a vector with components (# 44.1#, 33.1#, 66.1#, 11.1# #).

  • New instructions for minimum/maximum, such as minDouble# and minFloatX4#. These instructions compute the minimum/maximum of their inputs, working component-wise for SIMD vectors. Supported argument types are vector integer values (e.g. Word16X8#, Int32X4# etc) and both scalar and vector floating point values (e.g. Float#, DoubleX2#, FloatX8# etc).

2.1.8. ghc library

2.1.9. ghc-heap library

2.1.10. ghc-experimental library

2.1.11. template-haskell library

2.1.12. Included libraries

The package database provided with this distribution also contains a number of packages other than GHC itself. See the changelogs provided with these packages for further change information.

Package Version Reason for inclusion

ghc

 9.12

The compiler itself

Cabal-syntax

 3.14.0.0

Dependency of ghc-pkg utility

Cabal

 3.14.0.0

Dependency of ghc-pkg utility

Win32

 2.14.0.0

Dependency of ghc library

array

 0.5.8.0

Dependency of ghc library

base

 4.21.0.0

Core library

binary

 0.8.9.1

Dependency of ghc library

bytestring

 0.12.1.0

Dependency of ghc library

containers

 0.7

Dependency of ghc library

deepseq

 1.5.1.0

Dependency of ghc library

directory

 1.3.9.0

Dependency of ghc library

exceptions

 0.10.7

Dependency of ghc and haskeline library

file-io

 0.1.2

Dependency of directory library

filepath

 1.5.2.0

Dependency of ghc library

ghc-boot-th

 9.12

Internal compiler library

ghc-boot

 9.12

Internal compiler library

ghc-compact

 0.1.0.0

Core library

ghc-heap

 9.12

GHC heap-walking library

ghc-prim

 0.11.0

Core library

ghci

 9.12

The REPL interface

haskeline

 0.8.2.1

Dependency of ghci executable

hpc

 0.7.0.1

Dependency of hpc executable

integer-gmp

 1.1

Core library

mtl

 2.3.1

Dependency of Cabal library

os-string

 2.0.4

Dependency of filepath library

parsec

 3.1.17.0

Dependency of Cabal library

pretty

 1.1.3.6

Dependency of ghc library

process

 1.6.18.0

Dependency of ghc library

stm

 2.5.3.1

Dependency of haskeline library

template-haskell

 2.22.1.0

Core library

terminfo

 0.4.1.6

Dependency of haskeline library

text

 2.1.1

Dependency of Cabal library

time

 1.14

Dependency of ghc library

transformers

 0.6.1.1

Dependency of ghc library

unix

 2.8.5.1

Dependency of ghc library

xhtml

 3000.2.2.1

Dependency of haddock executable

haddock-api

 2.30.0

Dependency of haddock executable

haddock-library

 1.11.0

Dependency of haddock executable