Simon Peyton Jones pushed to branch wip/T23162-spj at Glasgow Haskell Compiler / GHC
Commits:
4038a28b by Andreas Klebinger at 2025-10-30T12:38:52-04:00
Add a perf test for #26425
- - - - -
f997618e by Andreas Klebinger at 2025-10-30T12:38:52-04:00
OccAnal: Be stricter for better compiler perf.
In particular we are now stricter:
* When combining usageDetails.
* When computing binder info.
In combineUsageDetails when combining the underlying adds we compute a
new `LocalOcc` for each entry by combining the two existing ones.
Rather than wait for those entries to be forced down the road we now
force them immediately. Speeding up T26425 by about 10% with little
effect on the common case.
We also force binders we put into the Core AST everywhere now.
Failure to do so risks leaking the occ env used to set the binders
OccInfo.
For T26425 compiler residency went down by a factor of ~10x.
Compile time also improved by a factor of ~1.6.
-------------------------
Metric Decrease:
T18698a
T26425
T9233
-------------------------
- - - - -
5618645b by Vladislav Zavialov at 2025-10-30T12:39:33-04:00
Fix namespace specifiers in subordinate exports (#12488)
This patch fixes an oversight in the `lookupChildrenExport` function that
caused explicit namespace specifiers of subordinate export items to be
ignored:
module M (T (type A)) where -- should be rejected
data T = A
Based on the `IEWrappedName` data type, there are 5 cases to consider:
1. Unadorned name: P(X)
2. Named default: P(default X)
3. Pattern synonym: P(pattern X)
4. Type name: P(type X)
5. Data name: P(data X)
Case 1 is already handled correctly; cases 2 and 3 are parse errors; and
it is cases 4 and 5 that we are concerned with in this patch.
Following the precedent established in `LookupExactName`, we introduce
a boolean flag in `LookupChildren` to control whether to look up in all
namespaces or in a specific one. If an export item is accompanied by an
explicit namespace specifier `type` or `data`, we restrict the lookup in
`lookupGRE` to a specific namespace.
The newly introduced diagnostic `TcRnExportedSubordinateNotFound`
provides error messages and suggestions more tailored to this context
than the previously used `reportUnboundName`.
- - - - -
2cc9924e by Richard Eisenberg at 2025-10-30T17:08:32+00:00
Refactor fundep solving
This commit is a large-scale refactor of the increasingly-messy code that
handles functional dependencies. It has virtually no effect on what compiles
but improves error messages a bit. And it does the groundwork for #23162.
The big picture is described in
Note [Overview of functional dependencies in type inference]
in GHC.Tc.Solver.FunDeps
* New module GHC.Tc.Solver.FunDeps contains all the fundep-handling
code for the constraint solver.
* Fundep-equalities are solved in a nested scope; they may generate
unifications but otherwise have no other effect.
See GHC.Tc.Solver.FunDeps.solveFunDeps
The nested needs to start from the Givens in the inert set, but
not the Wanteds; hence a new function `resetInertCans`, used in
`nestFunDepsTcS`.
* That in turn means that fundep equalities never show up in error
messages, so the complicated FunDepOrigin tracking can all disappear.
* We need to be careful about tracking unifications, so we kick out
constraints from the inert set after doing unifications. Unification
tracking has been majorly reformed: see Note [WhatUnifications] in
GHC.Tc.Utils.Unify.
A good consequence is that the hard-to-grok `resetUnificationFlag`
has been replaced with a simpler use of
`reportCoarseGrainUnifications`
Smaller things:
* Rename `FunDepEqn` to `FunDepEqns` since it contains multiple
type equalities.
Some compile time improvement
Metrics: compile_time/bytes allocated
Baseline
Test value New value Change
---------------------- --------------------------------------
T5030(normal) 173,839,232 148,115,248 -14.8% GOOD
hard_hole_fits(normal) 286,768,048 284,015,416 -1.0%
geo. mean -0.2%
minimum -14.8%
maximum +0.3%
Metric Decrease:
T5030
- - - - -
072466a4 by Simon Peyton Jones at 2025-10-30T17:08:32+00:00
QuickLook's tcInstFun should make instantiation variables directly
tcInstFun must make "instantiation variables", not regular
unification variables, when instantiating function types. That was
previously implemented by a hack: set the /ambient/ level to QLInstTyVar.
But the hack finally bit me, when I was refactoring WhatUnifications.
And it was always wrong: see the now-expunged (TCAPP2) note.
This commit does it right, by making tcInstFun call its own
instantiation functions. That entails a small bit of duplication,
but the result is much, much cleaner.
- - - - -
0c9e957c by Simon Peyton Jones at 2025-10-30T17:08:32+00:00
Build implication for constraints from (static e)
This commit addresses #26466, by buiding an implication for the
constraints arising from a (static e) form. The implication has
a special ic_info field of StaticFormSkol, which tells the constraint
solver to use an empty set of Givens.
See (SF3) in Note [Grand plan for static forms]
in GHC.Iface.Tidy.StaticPtrTable
This commit also reinstates an `assert` in GHC.Tc.Solver.Equality.
The test `StaticPtrTypeFamily` was failing with an assertion failure,
but it now works.
- - - - -
fbb2a831 by Simon Peyton Jones at 2025-10-30T17:08:32+00:00
Comments about defaulting representation equalities
- - - - -
b59a8713 by Simon Peyton Jones at 2025-10-30T17:08:32+00:00
Improve tracking of rewriter-sets
This refactor substantially improves the treatment of so-called
"rewriter-sets" in the constraint solver.
The story is described in the rewritten
Note [Wanteds rewrite Wanteds: rewriter-sets]
in GHC.Tc.Types.Constraint
Some highlights
* Trace the free coercion holes of a filled CoercionHole,
in CoercionPlusHoles. See Note [Coercion holes] (COH5)
This avoids taking having to take the free coercion variables
of a coercion when zonking a rewrriter-set
* Many knock on changes
* Make fillCoercionHole take CoercionPlusHoles as its argument
rather than to separate arguments.
* Similarly setEqIfWanted, setWantedE, wrapUnifierAndEmit.
* Be more careful about passing the correct CoHoleSet to
`rewriteEqEvidence` and friends
* Make kickOurAfterFillingCoercionHole more clever. See
new Note [Kick out after filling a coercion hole]
Smaller matters
* Rename RewriterSet to CoHoleSet
* Add special-case helper `rewriteEqEvidenceSwapOnly`
- - - - -
02de4352 by Simon Peyton Jones at 2025-10-30T17:08:32+00:00
Tidy up constraint solving for foralls
* In `can_eq_nc_forall` make sure to track Givens that are used
in the nested solve step.
* Tiny missing-swap bug-fix in `lookup_eq_in_qcis`
* Fix some leftover mess from
commit 14123ee646f2b9738a917b7cec30f9d3941c13de
Author: Simon Peyton Jones