If solution application is attempted for one of the sub-expressions,
while diagnostics are trying to narrow down where the failure is
located, don't record that captures need to be computed for closures,
because that could fail later on as in such conditions expressions
are not guaranteed to have correct types (e.g. some types could be
set to "unresolved").
Merging partial solutions can end up assigning the same type to a
particular typed node (expression, parameter, etc.), which can lead to
unbalanced set/clear when exploring the solution space (and later on,
crashes). Don't re-insert such information.
This is the same approach taken for type variable bindings, but it's
all pretty unfortunate: partial solutions should only record
information relative to their part of the constraint system, which
would save time and memory during solving. Howver, that's too big a
change for right now.
Fixes rdar://problem/50853028.
When we perform constraint generation while solving, capture the
type mappings we generate as part of the solver scope and solutions,
rolling them back and replaying them as necessary. Otherwise, we’ll
end up with uses of stale type variables, expression/parameter/type-loc
types set twice, etc.
Fixes rdar://problem/50390449 and rdar://problem/50150314.
Use the opened type from the callee declaration to open up references to
generic function builders that contain type parameters. This allows general
use of generic function builders.
When calling a function whose parameter specifies a function builder
with a multi-statement closure argument, transform the closure into
a single expression via the function builder. Should the result
type checker, replace the closure body with the single expression.
- Add a precondition on `doesDeclRefApplyCurriedSelf` to expect
a member decl, and rename it to make the precondition explicit.
- Don't assume that not having a base type means this isn't a member
reference, as member references to static operators don't have base
types.
Resolves SR-10843.
Use this function to replace various places where the logic is
duplicated.
In addition, isolate the logic where subscripts are treated as having
curried self parameters to CalleeCandidateInfo, as their interface types
don't have a curried self, but get curried with self by
CalleeCandidateInfo. Ideally we'd fix this by having a subscript's
interface type be curried with self, but given that most of this CSDiag
logic should be going away, this may not be necessary.
While computing a type of member via `getTypeOfMemberReference`
let's delay opening generic requirements associated with function
type until after self constraint has been created, that would give
a chance for contextual types to get propagated and make mismatch
originated in generic requirements much easier to diagnose.
Consider following example:
```swift
struct S<T> {}
extension S where T == Int {
func foo() {}
}
func test(_ s: S<String>) {
s.foo()
}
```
`foo` would get opened as `(S<$T>) -> () -> Void` and contextual `self`
type is going to be `S<String>`, so applying that before generic requirement
`$T == Int` would make sure that `$T` gets bound to a contextual
type of `String` and later fails requirement constraint `$T == Int`.
This is much easier to diagnose comparing to `$T` being bound to
`Int` right away due to same-type generic requirement and then
failing an attempt to convert `S<String>` to `S<Int>` while simplifying
self constraint.
Resolves: rdar://problem/46427500
Resolves: rdar://problem/34770265
Currently we check that the fixes share the same anchor, however this
doesn't account for the case where, given an ApplyExpr, one fix is
anchored on its function expr, and another is anchored on the apply
itself (because the former fix might be looking at the callee's
requirements, and the latter fix might be looking at an argument of
the call).
This commit changes the logic such that we check that fixes share the
same callee locator, which covers the above case. In addition, now
that we have the callee locator, we can use this to find the overload
directly.
This commit adds `ConstraintSystem::getCalleeLocator`, which forms a
locator that describes the callee of a given expression. This function
is then used to replace various places where this logic is duplicated.
This commit also changes the conditions under which a ConstructorMember
callee locator is formed. Previously it was formed for a CallExpr with a
TypeExpr function expr. However, now such a locator is formed if the
function expr is of AnyMetatypeType. This allows it to be more lenient
with invalid code, as well as work with DotSelfExpr.
Resolves SR-10694.
Introduce an attribute @_disfavoredOverload that can be used to state
that a particular declaration should be avoided if there is a
successful type-check for a non-@_disfavoredOverload. It's a way to
nudge overload resolution away from particular solutions.
The ConstraintSystem class is on the order of 1000s of bytes in size on
the stacka nd is causing issues with dispatch's 64k stack limit.
This changes most Small data types which store data on the stack to non
small heap based data types.
Instead of always requiring a call to be made to pass argument
to `@autoclosure` parameter, it should be allowed to pass argument
by value to `@autoclosure` parameter which can return a function
type.
```swift
func foo<T>(_ fn: @autoclosure () -> T) {}
func bar(_ fn: @autoclosure @escaping () -> Int) { foo(fn) }
```
Enhance call-argument matching to reject trailing closures that match up
with parameters that cannot accept closures at all.
Fixes rdar://problem/50362170.
Detect situations where key path doesn't have capability required
by the context e.g. read-only vs. writable, or either root or value
types are incorrect e.g.
```swift
struct S { let foo: Int }
let _: WritableKeyPath<S, Int> = \.foo
```
Here context requires a writable key path but `foo` property is
read-only.
Currently `getPotentialBindingsForRelationalConstraint` doesn't
respect the fact that type of key path expression has to be a
form of `KeyPath`, instead it could eagerly try to bind it to
`Any` or other contextual type if it's only available
information.
This patch aims to fix this situation by filtering potential
bindings available for type variable representing type of
the key path expression.
Resolves: SR-10467
We have a systemic class of issues where noescape types end up bound to
type variables in places that should not. The existing diagnostic for
this is ad-hoc and duplicated in several places but it doesn't actually
address the root cause of the problem.
For now, I've changed all call sites of createTypeVariable() to set the
new flag. I plan on removing enough occurrences of the flag to replicate
the old diagnostics. Then we can continue to refine this over time.
Implicit argument expression was necessary to generate keypath
constraint which is used to validate a choice picked for the member.
But since read-only check has been factored out it's now possible
to validate choice directly in combination with new 'keypath dynamic lookup'
locator associated with member type variable which represents result
of the dynamic lookup.
Instead of waiting until the overload is attempted, let's figure out
if there is anything wrong with it beforehand and attach a fix to the
"bind overload" constraint representing it.
