Vladislav Zavialov pushed to branch wip/int-index/visible-forall-gadts at Glasgow Haskell Compiler / GHC
Commits:
35826d8b by Matthew Pickering at 2025-06-08T22:00:41+01:00
Hadrian: Add option to generate .hie files for stage1 libraries
The +hie_files flavour transformer can be enabled to produce hie files
for stage1 libraries. The hie files are produced in the
"extra-compilation-artifacts" folder and copied into the resulting
bindist.
At the moment the hie files are not produced for the release flavour,
they add about 170M to the final bindist.
Towards #16901
- - - - -
e2467dbd by Ryan Hendrickson at 2025-06-09T13:07:05-04:00
Fix various failures to -fprint-unicode-syntax
- - - - -
1d99d3e4 by maralorn at 2025-06-12T03:47:39-04:00
Add necessary flag for js linking
- - - - -
974d5734 by maralorn at 2025-06-12T03:47:39-04:00
Don’t use additional linker flags to detect presence of -fno-pie in configure.ac
This mirrors the behavior of ghc-toolchain
- - - - -
1e9eb118 by Andrew Lelechenko at 2025-06-12T03:48:21-04:00
Add HasCallStack to Control.Monad.Fail.fail
CLC proposal https://github.com/haskell/core-libraries-committee/issues/327
2% compile-time allocations increase in T3064, likely because `fail`
is now marginally more expensive to compile.
Metric Increase:
T3064
- - - - -
6d12060f by meooow25 at 2025-06-12T14:26:07-04:00
Bump containers submodule to 0.8
Also
* Disable -Wunused-imports for containers
* Allow containers-0.8 for in-tree packages
* Bump some submodules so that they allow containers-0.8. These are not
at any particular versions.
* Remove unused deps containers and split from ucd2haskell
* Fix tests affected by the new containers and hpc-bin
- - - - -
537bd233 by Peng Fan at 2025-06-12T14:27:02-04:00
NCG/LA64: Optimize code generation and reduce build-directory size.
1. makeFarBranches: Prioritize fewer instruction sequences.
2. Prefer instructions with immediate numbers to reduce register moves,
e.g. andi,ori,xori,addi.
3. Ppr: Remove unnecessary judgments.
4. genJump: Avoid "ld+jr" as much as possible.
5. BCOND and BCOND1: Implement conditional jumps with two jump ranges,
with limited choice of the shortest.
6. Implement FSQRT, CLT, CTZ.
7. Remove unnecessary code.
- - - - -
19f20861 by Simon Peyton Jones at 2025-06-13T09:51:11-04:00
Improve redundant constraints for instance decls
Addresses #25992, which showed that the default methods
of an instance decl could make GHC fail to report redundant
constraints.
Figuring out how to do this led me to refactor the computation
of redundant constraints. See the entirely rewritten
Note [Tracking redundant constraints]
in GHC.Tc.Solver.Solve
- - - - -
1d02798e by Matthew Pickering at 2025-06-13T09:51:54-04:00
Refactor the treatment of nested Template Haskell splices
* The difference between a normal splice, a quasiquoter and implicit
splice caused by lifting is stored in the AST after renaming.
* Information that the renamer learns about splices is stored in the
relevant splice extension points (XUntypedSpliceExpr, XQuasiQuote).
* Normal splices and quasi quotes record the flavour of splice
(exp/pat/dec etc)
* Implicit lifting stores information about why the lift was attempted,
so if it fails, that can be reported to the user.
* After renaming, the decision taken to attempt to implicitly lift a
variable is stored in the `XXUntypedSplice` extension field in the
`HsImplicitLiftSplice` constructor.
* Since all the information is stored in the AST, in `HsUntypedSplice`,
the type of `PendingRnSplice` now just stores a `HsUntypedSplice`.
* Error messages since the original program can be easily
printed, this is noticeable in the case of implicit lifting.
* The user-written syntax is directly type-checked. Before, some
desugaring took place in the
* Fixes .hie files to work better with nested splices (nested splices
are not indexed)
* The location of the quoter in a quasiquote is now located, so error
messages will precisely point to it (and again, it is indexed by hie
files)
In the future, the typechecked AST should also retain information about
the splices and the specific desugaring being left to the desugarer.
Also, `runRnSplice` should call `tcUntypedSplice`, otherwise the
typechecking logic is duplicated (see the `QQError` and `QQTopError`
tests for a difference caused by this).
- - - - -
f93798ba by Cheng Shao at 2025-06-13T09:52:35-04:00
libffi: update to 3.5.1
Bumps libffi submodule.
- - - - -
a013284c by Vladislav Zavialov at 2025-06-13T18:24:15+03:00
Visible forall in GADTs
Add support for visible dependent quantification `forall a -> t` in
types of data constructors, e.g.
data KindVal a where
K :: forall k.
forall (a::k) -> -- now allowed!
k ->
KindVal a
For details, see docs/users_guide/exts/required_type_arguments.rst,
which has gained a new subsection.
DataCon in compiler/GHC/Core/DataCon.hs
---------------------------------------
The main change in this patch is that DataCon, the Core representation
of a data constructor, now uses a different type to store user-written
type variable binders:
- dcUserTyVarBinders :: [InvisTVBinder]
+ dcUserTyVarBinders :: [TyVarBinder]
where
type TyVarBinder = VarBndr TyVar ForAllTyFlag
type InvisTVBinder = VarBndr TyVar Specificity
and
data Specificity = InferredSpec | SpecifiedSpec
data ForAllTyFlag = Invisible Specificity | Required
This change necessitates some boring, mechanical changes scattered
throughout the diff:
... is now used in place of ...
-----------------+---------------
TyVarBinder | InvisTVBinder
IfaceForAllBndr | IfaceForAllSpecBndr
Specified | SpecifiedSpec
Inferred | InferredSpec
mkForAllTys | mkInvisForAllTys
additionally,
tyVarSpecToBinders -- added or removed calls
ifaceForAllSpecToBndrs -- removed calls
Visibility casts in mkDataConRep
--------------------------------
Type abstractions in Core (/\a. e) always have type (forall a. t)
because coreTyLamForAllTyFlag = Specified. This is also true of data
constructor workers. So we may be faced with the following:
data con worker: (forall a. blah)
data con wrapper: (forall a -> blah)
In this case the wrapper must use a visibility cast (e |> ForAllCo ...)
with appropriately set fco_vis{L,R}. Relevant functions:
mkDataConRep in compiler/GHC/Types/Id/Make.hs
dataConUserTyVarBindersNeedWrapper in compiler/GHC/Core/DataCon.hs
mkForAllVisCos in compiler/GHC/Core/Coercion.hs
mkCoreTyLams in compiler/GHC/Core/Make.hs
mkWpForAllCast in compiler/GHC/Tc/Types/Evidence.hs
More specifically:
- dataConUserTyVarBindersNeedWrapper has been updated to answer "yes"
if there are visible foralls in the type of the data constructor.
- mkDataConRep now uses mkCoreTyLams to generate the big lambda
abstractions (/\a b c. e) in the data con wrapper.
- mkCoreTyLams is a variant of mkCoreLams that applies visibility casts
as needed. It similar in purpose to the pre-existing mkWpForAllCast,
so the common bits have been factored out into mkForAllVisCos.
ConDecl in compiler/Language/Haskell/Syntax/Decls.hs
----------------------------------------------------
The surface syntax representation of a data constructor declaration is
ConDecl. In accordance with the proposal, only GADT syntax is extended
with support for visible forall, so we are interested in ConDeclGADT.
ConDeclGADT's field con_bndrs has been renamed to con_outer_bndrs
and is now accompanied by con_inner_bndrs:
con_outer_bndrs :: XRec pass (HsOuterSigTyVarBndrs pass)
con_inner_bndrs :: [HsForAllTelescope pass]
Visible foralls always end up in con_inner_bndrs. The outer binders are
stored and processed separately to support implicit quantification and
the forall-or-nothing rule, a design established by HsSigType.
A side effect of this change is that even in absence of visible foralls,
GHC now permits multiple invisible foralls, e.g.
data T a where { MkT :: forall a b. forall c d. ... -> T a }
But of course, this is done in service of making at least some of these
foralls visible. The entire compiler front-end has been updated to deal
with con_inner_bndrs. See the following modified or added functions:
Parser:
mkGadtDecl in compiler/GHC/Parser/PostProcess.hs
splitLHsGadtTy in compiler/GHC/Hs/Type.hs
Pretty-printer:
pprConDecl in compiler/GHC/Hs/Decls.hs
pprHsForAllTelescope in compiler/GHC/Hs/Type.hs
Renamer:
rnConDecl in compiler/GHC/Rename/Module.hs
bindHsForAllTelescopes in compiler/GHC/Rename/HsType.hs
extractHsForAllTelescopes in compiler/GHC/Rename/HsType.hs
Type checker:
tcConDecl in compiler/GHC/Tc/TyCl.hs
tcGadtConTyVarBndrs in compiler/GHC/Tc/Gen/HsType.hs
Template Haskell
----------------
The TH AST is left unchanged for the moment to avoid breakage. An
attempt to quote or reify a data constructor declaration with visible
forall in its type will result an error:
data ThRejectionReason -- in GHC/HsToCore/Errors/Types.hs
= ...
| ThDataConVisibleForall -- new error constructor
However, as noted in the previous section, GHC now permits multiple
invisible foralls, and TH was updated accordingly. Updated code:
repC in compiler/GHC/HsToCore/Quote.hs
reifyDataCon in compiler/GHC/Tc/Gen/Splice.hs
ppr @Con in libraries/ghc-boot-th/GHC/Boot/TH/Ppr.hs
Pattern matching
----------------
Everything described above concerns data constructor declarations, but
what about their use sites? Now it is trickier to type check a pattern
match fn(Con a b c)=... because we can no longer assume that a,b,c are
all value arguments. Indeed, some or all of them may very well turn out
to be required type arguments.
To that end, see the changes to:
tcDataConPat in compiler/GHC/Tc/Gen/Pat.hs
splitConTyArgs in compiler/GHC/Tc/Gen/Pat.hs
and the new helpers split_con_ty_args, zip_pats_bndrs.
This is also the reason the TcRnTooManyTyArgsInConPattern error
constructor has been removed. The new code emits TcRnArityMismatch
or TcRnIllegalInvisibleTypePattern.
Summary
-------
DataCon, ConDecl, as well as all related functions have been updated to
support required type arguments in data constructors.
Test cases:
HieGadtConSigs GadtConSigs_th_dump1 GadtConSigs_th_pprint1
T25127_data T25127_data_inst T25127_infix
T25127_newtype T25127_fail_th_quote T25127_fail_arity
TyAppPat_Tricky
Co-authored-by: mniip