Commit Graph

651 Commits

Author SHA1 Message Date
Slava Pestov
c47ff06a72 Sema: Tiny cleanup for matchTypes() 2025-01-27 16:42:03 -05:00
Doug Gregor
b3f2f00588 Suggest both @unsafe and @safe Fix-Its for unsafe types in signature 2025-01-23 07:47:19 -08:00
kntk
bdfb6099b2 [ConstraintSystem] Add index value (as an impact of the score kind) to output of debug constraints. (#78740)
* [ConstraintSystem] Add index value (as an impact of the score kind) to debug constraints.

* Use weight instead of index, change label 'value' -> 'impact'

* Fix debug constraints output format: remove 'components' and add 'weight' label
2025-01-22 17:52:56 -08:00
Slava Pestov
d1d2411bf3 Merge pull request #78267 from slavapestov/incremental-bindings-part-0
Sema: Some small cleanups in CSBindings.cpp and related code
2025-01-16 09:03:32 -05:00
Slava Pestov
1aff26eaa6 Sema: Store BindingSet inside the ConstraintGraphNode
Building the DenseMap in determineBestBindings() is extremely
expensive.

Also rename getCurrentBindings() to getPotentialBindings().
2025-01-15 22:30:32 -05:00
Anthony Latsis
87cf1179ff [NFC] Sema: Add Solution::getCalleeOverloadChoice 2025-01-16 02:35:26 +00:00
Slava Pestov
c12e0a1e8f Merge pull request #78470 from slavapestov/fix-rdar141961300
Sema: Fix local property wrappers on constructor
2025-01-08 21:33:12 -05:00
Slava Pestov
c17044381e Sema: Fix local property wrappers on constructor
Fixes rdar://problem/142443421.
2025-01-08 14:31:12 -05:00
Slava Pestov
cc2048764a Sema: Clean up local property wrapper bookkeeping 2025-01-08 14:31:12 -05:00
Pavel Yaskevich
8f7e71aa97 Merge pull request #78487 from xedin/Sendable-to-Any-for-lvalues
[CSApply] Sendable-to-Any: Add support for l-value to l-value and inout unsafe casts
2025-01-08 09:57:02 -08:00
Pavel Yaskevich
79f6b07961 [CSApply] Mark self parameter as inout when base/self match on deep equality
If type equality check fails we need to check whether the types
are the same with deep equality restriction since `any Sendable`
to `Any` conversion is now supported in generic argument positions
of @preconcurrency declarations. i.e. referencing a member on
`[any Sendable]` if member declared in an extension that expects
`Element` to be equal to `Any`.
2025-01-07 10:55:04 -08:00
Slava Pestov
2d17294d73 Merge pull request #78301 from slavapestov/remove-one-way-constraints
Sema: Remove ConstraintKind::OneWayBindParam and ConstraintKind::OneWayEqual
2025-01-05 10:38:14 -05:00
Anthony Latsis
5c24df22a0 Merge pull request #78363 from AnthonyLatsis/dicksonia-arborescens
[NFC] Sema: Tidy up `canOpenExistentialCallArgument`
2025-01-02 15:38:27 +00:00
Hamish Knight
9fb064ad2c [CS] Remove resolveInterfaceType
Use `simplifyType` instead with the new parameter
for getting an interface type. Also avoid using
`resolveInterfaceType` in CSApply since we don't
need the opened generic parameter mapping behavior.
2024-12-31 18:39:20 +00:00
Hamish Knight
9fe4abc6fd [CS] Simplify Solution::resolveInterfaceType
Rather than attempting to re-implement `simplifyType`,
tweak `Solution::simplifyType` such that it can
map the resulting type out of context, and can
turn type variables into their opened generic
parameters.
2024-12-31 18:39:19 +00:00
Anthony Latsis
fc1ad79ddd [NFC] ConstraintSystem: Rename openExistentialType to openAnyExistentialType
This method can open an existential metatype too, and we already use
`isAnyExistentialType` to mean "existential type or existential
metatype"
2024-12-24 17:42:54 +00:00
Slava Pestov
385d66f24e Sema: Remove ConstraintKind::OneWayEqual 2024-12-21 00:42:13 -08:00
Slava Pestov
5071e96a78 Sema: Don't generate OneWayEqual constraints for pattern bindings 2024-12-21 00:42:13 -08:00
Slava Pestov
0c128e5db7 Sema: Optimize ConstraintGraph::computeConnectedComponents()
Instead of starting a depth-first search from each type variable
and marking all type variables that haven't been marked yet,
we can implement this as a union-find.

We can also store the temporary state directly inside the
TypeVariableType::Implementation, instead of creating large
DenseMaps whose keys range over all type variables.
2024-12-18 17:30:43 -05:00
Slava Pestov
45346c8389 Sema: Remove ConstraintGraph::TypeVariables 2024-12-18 17:30:34 -05:00
Slava Pestov
c4e466dae7 Sema: Remove needless indirection 2024-12-18 17:30:22 -05:00
Pavel Yaskevich
8bd288447f [CSGen] NFC: Remove obsolete ConstraintSystem::{get, set}FavoredType 2024-12-17 11:36:41 -08:00
Pavel Yaskevich
8d5cb112ef [ConstraintSystem] Narrowly disable tryOptimizeGenericDisjunction when some of the arguments are number literals
Don't attempt this optimization if call has number literals.
This is intended to narrowly fix situations like:

```swift
func test<T: FloatingPoint>(_: T) { ... }
func test<T: Numeric>(_: T) { ... }

test(42)
```

The call should use `<T: Numeric>` overload even though the
`<T: FloatingPoint>` is a more specialized version because
selecting `<T: Numeric>` doesn't introduce non-default literal
types.
2024-12-17 11:36:40 -08:00
Pavel Yaskevich
f2a6677a6d [CSOptimizer] Infer argument candidates from calls to Double and CGFloat constructors
Helps situations like `1 + {Double, CGFloat}(...)` by inferring
a type for the second operand of `+` based on a type being constructed.

Currently limited to Double and CGFloat only since we need to
support implicit `Double<->CGFloat` conversion.

Helps situations like `1 + CGFloat(...)` by inferring a type for
the second operand of `+` based on a type being constructed.
Currently limited to known integer, floating-point and CGFloat types
since we know how they are structured.
2024-12-17 11:36:40 -08:00
Pavel Yaskevich
c429f5b9ec [CSOptimizer] NFC: Switch from llvm::Optional to std::optional post-rebase 2024-12-17 11:36:39 -08:00
Pavel Yaskevich
0fc6806922 [CSOptimizer] NFC: Switch to llvm::Optional 2024-12-17 11:36:39 -08:00
Pavel Yaskevich
e404ed722a [CSStep] Don't favor choices until the disjunction is picked 2024-12-17 11:36:38 -08:00
Pavel Yaskevich
4432c51f57 [CSGen] Remove ConstraintOptimizer and all favoring logic 2024-12-17 11:36:37 -08:00
Pavel Yaskevich
757ca24e8a [ConstraintSystem] Remove shrink 2024-12-17 11:36:37 -08:00
Slava Pestov
c0787f337f Sema: Fix soundness hole with variable initializers and opaque return types
Just because the type of the initializer expression is an opaque return type,
does not mean it is the opaque return type *for the variable being initialized*.

It looks like there is a bit of duplicated logic and layering violations going
on so I only fixed one caller of openOpaqueType(). This addresses the test case
in the issue. For the remaining calls I added FIXMEs to investigate what is
going on.

Fixes https://github.com/swiftlang/swift/issues/73245.
Fixes rdar://127180656.
2024-12-04 15:13:22 -05:00
Pavel Yaskevich
08720eef82 Merge pull request #77936 from xedin/revert-void-favoring
Revert changes related to `Void` binding favoring
2024-12-04 10:24:20 -08:00
Pavel Yaskevich
8aa6280bf3 Revert "[CSBindings] Don't favor application result types before application happens"
This reverts commit bc949c3680.
2024-12-03 13:15:46 -08:00
Hamish Knight
d25bdfa4a3 [CS] Remove hack for rdar://139234188
Now that "is compound" is a separate bit in
FunctionRefInfo, we can correctly track the
application level for an EnumElementPattern.
2024-12-03 11:32:11 +00:00
Hamish Knight
0fdb038fc9 Merge pull request #77896 from hamishknight/functionref-split
[AST] Split out "is compound" bit on FunctionRefKind
2024-12-03 09:53:37 +00:00
Pavel Yaskevich
8a0a083a45 Merge pull request #77797 from xedin/locatable-types
[AST/Sema]  Introduce a new type that has associated location in source
2024-12-02 16:33:43 -08:00
Hamish Knight
a4d51419ba [AST] NFC: Rename FunctionRefKind -> FunctionRefInfo 2024-12-02 14:11:32 +00:00
Hamish Knight
746135b4d7 [CS] Add a narrow hack for rdar://139234188
Currently we set `FunctionRefKind::Compound` for
enum element patterns with tuple sub-patterns to
ensure the member has argument labels stripped. As
such, we need to account for the correct application
level in `getNumApplications`. We ought to be
setting the correct FunctionRefKind and properly
handling the label matching in the solver though.
We also ought to consider changing FunctionRefKind
such that "is compound" is a separate bit from the
application level.

rdar://139234188
2024-11-22 13:43:58 +00:00
Pavel Yaskevich
e61bf2089c [ConstraintSystem] Make types of implicit result builder variables locatable
This would aid `FailureDiagnostic` in anchoring conformance failures to `View`.
2024-11-21 14:12:57 -08:00
Slava Pestov
25252cd683 Merge pull request #77728 from slavapestov/cg-cleanup
Sema: Small ConstraintGraph cleanup
2024-11-20 19:43:42 -05:00
Slava Pestov
7fef225c9f Sema: Allow turning off expression timer with -solver-expression-time-threshold=0 2024-11-20 13:46:19 -05:00
Slava Pestov
c8415aeac4 Sema: Record the trail step count in solver statistics
Also introduce two new frontend flags:

The -solver-scope-threshold flag sets the maximum number of scopes, which was
previously hardcoded to 1 million.

The -solver-trail-threshold flag sets the maximum number of trail steps,
which defaults to 64 million.
2024-11-20 13:46:19 -05:00
Slava Pestov
60d34a4162 Sema: Micro-optimize PotentialBindings
PotentialBindings is part of ConstraintGraphNode and there's no need
to store the ConstraintSystem and TypeVariableType twice.

Also it doesn't need to be optional either, because we no longer need
to reset and recompute bindings.
2024-11-20 13:35:22 -05:00
Slava Pestov
1a18d70f90 Sema: Remove ConstraintSystem::SolverState::isRecordingChanges()
All but two remaining call sites can be changed to just check for a
non-null solverState, because we want to assert if we're inside
of an active undo.

The two places inside binding inference can check isUndoActive()
directly.
2024-11-20 13:35:22 -05:00
Slava Pestov
de731b4d5d Sema: Preserve insertion order in Solution::typeBindings 2024-11-20 11:47:07 -05:00
Slava Pestov
a7b1e7854c Merge pull request #77174 from slavapestov/fix-undo-order
Sema: Undo changes in chronological order in SolverTrail::undo()
2024-11-20 08:00:53 -05:00
Slava Pestov
1a60685292 Sema: Address the FIXME in ConstraintGraph::mergeNodes() 2024-11-19 21:24:12 -05:00
Slava Pestov
1ecba36e76 Sema: Cache Solution::getTotalMemory()
This is showing up in performance profiles.
2024-11-19 16:04:57 -05:00
Slava Pestov
09df48f9fb Sema: Remove DenseMap usage when opening types 2024-11-18 11:40:38 -05:00
Slava Pestov
47156e006b AST: Introduce ProtocolConformanceRef::forAbstract() 2024-11-16 16:16:06 -05:00
Pavel Yaskevich
bc949c3680 [CSBindings] Don't favor application result types before application happens
Until `ApplicableFunction` constraint is simplified result type
associated with it cannot be bound because the binding set if
incomplete.

Resolves: rdar://139237088
2024-11-08 14:00:01 -08:00