Commit Graph

1087 Commits

Author SHA1 Message Date
Anthony Latsis
c63b737e92 Collapse all indirect equivalents to ValueDecl::getBaseIdentifier 2020-03-29 00:36:01 +03:00
Doug Gregor
85762c30f5 Merge pull request #30663 from DougGregor/property-wrapper-default-init-infer-type
[Property wrappers] Allow default initialization to infer a type.
2020-03-27 13:46:47 -07:00
Pavel Yaskevich
c99c9b95d0 [ConstraintSystem] Switch availability check to use correct API
Use `TypeChecker::checkDeclarationAvailability` instead of `isDeclAvailable`,
that is a proper API endpoint which does flag checking before calling
`isDeclAvailable` internally.
2020-03-27 00:43:09 -07:00
Pavel Yaskevich
81dbec09fb [ConstraintSystem] Don't check -disable-availability-checking
Checking for `-disable-availability-checking` in
`ConstraintSystem::isDeclUnavailable` caused a regression with
obsolete/introduced checking. Let's rely on
`DeclAttributes::isUnavailable` and `TypeChecker::isDeclAvailable`
to do the right thing instead.

Resolves: rdar://problem/60898369
2020-03-26 16:13:32 -07:00
Doug Gregor
d282e3c1e9 [Property wrappers] NFC rename initialValue -> wrappedValue where appropriate
wrappedValue is the new name, and we only support initialValue for backward
compatibility with earlier revisions of the property wrappers proposal.
2020-03-26 15:16:28 -07:00
Doug Gregor
5ac8d769fd [Property wrappers] Allow default initialization to infer a type.
Property wrappers are allowed to infer the type of a variable, but this
only worked when the property wrapper was provided with an explicit
initialization, e.g.,

    @WrapsAnInt() var x   // infers type Int from WrapsAnInt.wrappedValue

However, when default initialization is supported by the property wrapper,
dropping the parentheses would produce an error about the missing type
annotation

    @WrapsAnInt var x

Make this second case behave like the first, so that default initialization
works consistently with the explicitly-specified version.
Fixes rdar://problem/59471019.
2020-03-26 13:44:08 -07:00
Pavel Yaskevich
4b3d075fb7 [ConstraintSystem] Produce fallback diagnostic on mismatch between fixes/holes
If score indicates that a solution should have at least one fix
but it doesn't let's attempt to produce a fallback diagnostic inline.
2020-03-24 16:51:44 -07:00
Hamish Knight
c74a7512f7 [CS] NFC: Remove OverloadChoiceKind::BaseType
This doesn't appear to be used any more.
2020-03-22 22:51:58 -07:00
Pavel Yaskevich
d673ed0989 Merge pull request #30519 from xedin/rdar-52204414
[ConstraintSystem] Don't allow explicit closure result to be implicitly converted to `Void`
2020-03-20 09:36:17 -07:00
Pavel Yaskevich
89933c67da [ConstraintSystem] Don't allow explicit closure result to be implicitly converted to Void
It's allowed to convert a single statement closure from `(...) -> T` to `(...) -> Void`
_only_ if there is no explicit `return` in the body.

Resolves: [SR-12277](https://bugs.swift.org/browse/SR-12277)
Resolves: rdar://problem/52204414
2020-03-19 16:59:17 -07:00
Doug Gregor
3abea6be65 [Constraint system] Use the PatternBindingDecl context when possible.
Make sure that we're resolving types and patterns using the
PatternBindingDecl context, both for the type resolver context and the
contextual pattern used for pattern resolution.

Fixes a regression with implicitly-unwrapped options reported as
SR-11998 / rdar://problem/58455441.
2020-03-18 21:09:06 -07:00
Pavel Yaskevich
2d6170b847 [TypeChecker] Requestify check to determine if closure has explicit result
Determine whether closure body has any explicit `return`
statements which could produce a non-void result.
2020-03-18 17:06:15 -07:00
Pavel Yaskevich
2dd74706c9 Merge pull request #30411 from xedin/diags-to-use-solution-instead-of-cs
[Diagnostics] Switch `FailureDiagnostic` to use a solution
2020-03-13 20:20:58 -07:00
Pavel Yaskevich
3ab8710329 [Diagnostics] Cleanup use of constraint system by ConstraintFix/FailureDiagnostic 2020-03-13 15:02:10 -07:00
Pavel Yaskevich
0fc9d1b84a [Diagnostics] Move getFunctionArgApplyInfo to Solution
Instead of relying on constraint system having solution applied
let's pass a solution directly to `getFunctionArgApplyInfo`.
2020-03-13 12:46:25 -07:00
Robert Widmann
4e67eec878 [CS] Move getUnopenedTypeOfReference out of TypeChecker
For some reason, the changed caller in CS wasn't actually going to use any of the types in the constraint system from the entrypoint it was calling. Switch over to using the constraint-system-based entrypoint so we can pick up expression types consistently. Then, move the TypeChecker entrypoint onto ConstraintSystem to reduce the duplication here.

The remaining callers of buildCheckedRefExpr should be migrated.
2020-03-11 11:14:10 -07:00
Pavel Yaskevich
e86558fe4f Merge pull request #30277 from LucianoPAlmeida/SR-12019-dynamically-call-generic-constraint
[SR-12019] Fix assert when dynamicallyCall parameter does not satisfy generic constraint
2020-03-10 09:12:33 -07:00
Pavel Yaskevich
4a06b13abf [ConstraintSystem] Diagnose argument conflicts only for "representative" generic parameters
Consider following example:

```swift
struct MyType<TyA, TyB> {
  var a : TyA, b : TyB
}

typealias B<T1> = MyType<T1, T1>

_ = B(a: "foo", b: 42)
```

Here `T1` is equal to `TyA` and `TyB` so diagnostic about
conflicting arguments ('String' vs. 'Int') should only be
produced for "representative" `T1`.
2020-03-09 12:10:29 -07:00
Luciano Almeida
f796e27b65 [TypeChecker] Requestify DynamicCallableAttribute check 2020-03-08 20:44:23 -03:00
Pavel Yaskevich
0ecedfa5ea Revert "[ConstraintSystem] Make it possible to infer subtype bindings through argument conversions"
Reverts apple/swift#30006. It caused a regression that we'd like to address before re-landing:

```swift
struct X {
  var cgf: CGFloat
}

func test(x: X?) {
  let _ = (x?.cgf ?? 0) <= 0.5
}
```

This reverts commit 0a6b444b49.
This reverts commit ed255596a6.
This reverts commit 3e01160a2f.
This reverts commit 96297b7e39.

Resolves: rdar://problem/60185506
2020-03-07 20:16:56 -08:00
Luciano Almeida
6df132de1a [ConstraintSystem] Make getCalleLocator to support an apply involving an dynamicCallable 2020-03-07 16:43:37 -03:00
Pavel Yaskevich
7c19d1ccd1 [ConstraintSystem] Remove baseCS since CSDiag has been removed
There is no more sub-expression re-typechecking so no need to keep
parent constraint systems around.
2020-03-06 09:35:52 -08:00
Pavel Yaskevich
4c600843d6 [ConstraintSystem] Check availability as part of resolving overload choice
This is a follow up to changes related to contextual availability
(https://github.com/apple/swift/pull/29921) which increased score
for unavailable declarations only if they were overloaded but
overlooked a case of a single unavailable choice.

Resolve: rdar://problem/60047439
2020-03-05 00:35:47 -08:00
Holly Borla
fce8738ea8 [ConstraintSystem] Replace dependent member types with holes when the base type
doesn't conform to the associatedtype's protocol (only in diagnostic mode).

This allows the solver to find solutions for more cases involving requirement
failures for dependent member types without special cases across the solver
that check for dependent member types with no type variables.
2020-03-03 15:45:24 -08:00
Robert Widmann
d494cc8dcb Merge pull request #30109 from CodaFi/lies-more-lies-and-statistics
[Frontend] Clean Up Usage of UnifiedStatsReporter
2020-02-28 10:24:02 -08:00
Holly Borla
87bb7755c2 Merge pull request #30101 from hborla/dynamic-replacement-type-erasure
[Sema] Implement type erasure for dynamic replacement.
2020-02-28 09:37:33 -08:00
Robert Widmann
de72824b04 [Gardening] Canonicalize usages of ASTContext::Stats 2020-02-27 17:12:58 -08:00
Slava Pestov
019452f9af Sema: Diagnose unbound method references on 'super.'
This is something I noticed by inspection while working on
<https://bugs.swift.org/browse/SR-75>.

Inside a static method, 'self' is a metatype value, so
'self.instanceMethod' produces an unbound reference of type
(Self) -> (Args...) -> Results.

You might guess that 'super.instanceMethod' can similarly
be used to produce an unbound method reference that calls
the superclass method given any 'self' value, but unfortunately
it doesn't work.

Instead, 'super.instanceMethod' would produce the same
result as 'self.instanceMethod'. Maybe we can implement this
later, but for now, let's just diagnose the problem.

Note that partially-applied method references with 'super.'
-- namely, 'self.staticMethod' inside a static context, or
'self.instanceMethod' inside an instance context, continue
to work as before.

They have the type (Args...) -> Result; since the self value
has already been applied we don't hit the representational
issue.
2020-02-27 17:28:23 -05:00
Holly Borla
3cdc30ffeb [Sema] Support type erasure for dynamic replacement in function
builders.
2020-02-27 09:01:16 -08:00
Slava Pestov
90ee606de7 Sema: Refactor CSApply in preparation for curry thunks 2020-02-26 23:10:07 -05:00
Doug Gregor
cc8db01ba9 Merge pull request #30045 from DougGregor/function-builders-if-let
[Constraint system] Support if let / if case in function builders.
2020-02-25 16:24:09 -08:00
Doug Gregor
3f2f79a699 [Constraint system] Fold pattern variable binding into constraint gen.
Rather than re-walk the pattern to create type bindings for the variables
that show up in the pattern, assign types to each of the variables as part
of constraint generation for the pattern. Only do this in contexts
where we will need the types, e.g., function builders.
2020-02-25 08:51:56 -08:00
Pavel Yaskevich
0a6b444b49 [ConstraintSystem] Diagnose argument conflicts only for "representative" generic parameters
Consider following example:

```swift
struct MyType<TyA, TyB> {
  var a : TyA, b : TyB
}

typealias B<T1> = MyType<T1, T1>

_ = B(a: "foo", b: 42)
```

Here `T1` is equal to `TyA` and `TyB` so diagnostic about
conflicting arguments ('String' vs. 'Int') should only be
produced for "representative" `T1`.
2020-02-21 16:49:37 -08:00
Pavel Yaskevich
71753f3ca6 [ConstraintSystem] Rank contextually unavailable overloads lower than other choices (#29921)
Currently constraint solver is only capable of detecting universally unavailable
overloads but that's insufficient because it's still possible to pick a contextually
unavailable overload choice which could be better than e.g. generic overload, or
one with defaulted arguments, marked as disfavored etc.

Let's introduce `ConstraintSystem::isDeclUnavailable` which supports both universal
and contextual unavailability and allow constraint solver to rank all unavailable
overload choices lower than any other possible choice(s).

Resolves: rdar://problem/59056638
2020-02-19 13:13:53 -05:00
Hamish Knight
ed1d372887 [CS] Remove DiagnosedExprs 2020-02-19 07:43:59 -08:00
Pavel Yaskevich
2c92df4a6d [ConstraintSystem] Simplify diagnoseAmbiguityWithEphemeralPointers
Don't attempt to figure out what exactly is ambiguous, let
`diagnoseAmbiguity` take care of that. Simplify make sure
that only some of the solutions have fixes and these fixes
are all related to use of ephemeral pointers.
2020-02-18 15:06:13 -08:00
Pavel Yaskevich
2dccdbfabf [ConstraintSystem] NFC: Remove workarounds related to (now deprecated) CSDiag 2020-02-18 10:13:09 -08:00
Pavel Yaskevich
b905113f5c [ConstraintSystem] Remove now completely obsolete CSDiag 2020-02-18 09:12:56 -08:00
Pavel Yaskevich
0c451bb2a9 [Diagnostics] Diagnose ambiguities related to ephemeral pointers
Diagnose ambiguity related to overloaded declarations where only
*some* of the overload choices have ephemeral pointer warnings/errors
associated with them. Such situations have be handled specifically
because ephemeral fixes do not affect the score.

If all of the overloads have ephemeral fixes associated with them
it's much easier to diagnose through notes associated with each fix.
2020-02-17 16:09:11 -08:00
Luciano Almeida
a32e3f4598 [docs] Updating TypeChecker.rst 2020-02-16 18:41:57 -03:00
Luciano Almeida
6c617f9718 [ConstraintSystem] Removing check related to ArgumentMismatch in CSDiag. 2020-02-16 16:51:41 -03:00
Puyan Lotfi
b707b7f0b3 Fixing build breaks for master-next against latest llvm.org.
There were changes due to the StringRef to std::string conversion, changes
in the Debug Info DIBuilder::createModule API, and a drop in the using for
PointerUnion4 since PointerUnion is now a variadic template and will do in its
place.
2020-02-13 14:12:21 -08:00
Pavel Yaskevich
355148dbe8 Merge pull request #29746 from hborla/ambiguity-with-fixes
[Diagnostics] Refactor diagnoseAmbiguityWithFixes.
2020-02-13 02:42:43 -08:00
Pavel Yaskevich
172c2003ea [Diagnostics] NFC: Add a comment about special handling of opaque result types 2020-02-13 00:54:51 -08:00
Pavel Yaskevich
b872bdfca1 [Diagnostics] Replace llvm::find_if with llvm::any_of in diagnoseAmbiguityWithFixes 2020-02-13 00:45:44 -08:00
Pavel Yaskevich
60a210b657 [ConstraintSystem] NFC: Fix a couple of typos in comments 2020-02-13 00:44:20 -08:00
Pavel Yaskevich
3f10bfa74e [Diagnostics] NFC: Rename diagnoseConflictingArguments into diagnoseConflictingGenericArguments 2020-02-12 18:17:10 -08:00
Pavel Yaskevich
af65c0e2f3 [Diagnostics] Prioritize contextual type mismatch other ambiguities 2020-02-12 17:55:32 -08:00
Pavel Yaskevich
5437622d2d [Diagnostics] Diagnose ambiguity with conflicting arguments to generic parameters
It's done by first retrieving all generic parameters from each solution,
filtering boundings into distrinct set and diagnosing any differences.

For example:

```swift
func foo<T>(_: T, _: T) {}
func bar(x: Int, y: Float) {
  foo(x, y)
}
```
2020-02-12 17:53:31 -08:00
Robert Widmann
054d7b9913 Merge pull request #29713 from CodaFi/unused-unwanted-unloved
[Gardening] Silence Some Warning Spew
2020-02-11 15:46:16 -08:00