To better preserve source compatibility, teach the constraint
solver to try both the new forward scanning rule as well as the
backward scanning rule when matching a single, unlabeled trailing
closure. In the extreme case, where the unlabeled trailing closure
matches different parameters with the different rules, and yet both
produce a potential match, introduce a disjunction to explore both
possibilities.
Prefer solutions that involve forward scans to those that involve
backward scans, so we only use the backward scan as a fallback.
After the TypeLocs were removed here, the TypeRepr from the IsExpr was
the only thing providing access to syntactic information from the parent
IsExpr. In order to support this, it was possible to construct a bizarre
ConditionalCheckedCastExpr that contained both semantic and syntactic
information. This doesn't comport with the rest of the casting nodes,
which force you to pick one or the other.
Since we're rewriting an IsExpr into a EnumIsCaseExpr, let's just stash
the syntactic information there. This unblocks a bit of cleanup.
Reverse the polarity of the "checked in context" bit for ClosureExpr
to "separately checked", which simplifies the AST walker logic (to
"should we walk separately type-checked closure bodies?") and
eliminates single-expression closures as a separate case to consider.
Rather than using various "applied function builder" and "is single
expression body" checks to determine whether a closure was
type-checked in its enclosing expression, record in the closure
expression whether it actually *was* type-checked as part of its
enclosing expression.
Introduce a statement visitor that applies a particular solution to
the body of a closure. This matches the mechanism used by function
builders (and is similar to how we handle expressions in general),
simplifying the logic for handling
conversion-to-void-returning-closures and
conversion-from-Never-returning-bodies. It is a stepping stone for
type inference of multi-statement closures.
This restores getSourceRange() on DefaultArgumentExpr after it was removed in
https://github.com/apple/swift/pull/31184.
It was originally removed to solve the issues it was causing when computing the
source range of its parent TupleExpr. To account for trailing closures we walk
back through the tuple's arguments until one with a valid location is found,
which we use as the end location. If the last argument was a DefaultArgumentExpr
though that meant the end loc would end up being the tuple's start location, so
none of the tuple's other arguments were contained in its range, triggering an
ASTVerifier assertion. Source tooling and diagnostics don't care about default
arg expression locations as nothing can reference them, but their locations are
output in the debug info. Added a regression test to catch that in future, and
updated TupleExpr::getSourceRange() to ignore them when computing the end loc.
Resolves rdar://problem/63195504.
that allows arbitrary `label: {}` suffixes after an initial
unlabeled closure.
Type-checking is not yet correct, as well as code-completion
and other kinds of tooling.
Accept trailing closures in following form:
```swift
foo {
<label-1>: { ... }
<label-2>: { ... }
...
<label-N>: { ... }
}
```
Consider each labeled block to be a regular argument to a call or subscript,
so the result of parser looks like this:
```swift
foo(<label-1>: { ... }, ..., <label-N>: { ... })
```
Note that in this example parens surrounding parameter list are implicit
and for the cases when they are given by the user e.g.
```swift
foo(bar) {
<label-1>: { ... }
...
}
```
location of `)` is changed to a location of `}` to make sure that call
"covers" all of the transformed arguments and parser result would look
like this:
```swift
foo(bar,
<label-1>: { ... }
)
```
Resolves: rdar://problem/59203764
Also extend returned object from simplify being an expression to
`TrailingClosure` which has a label, label's source location and
associated closure expression.
For DoubleCurryThunk cases it’s expecting an ApplyExpr directly within the
OpenExistentialExpr, but in some cases it contains an ErasureExpr (implicit
conversion) that wraps the ApplyExpr. This updates the method to look
through implicit conversions.
Resolves rdar://problem/61885996
Remove duplication in the modeling of TypeExpr. The type of a TypeExpr
node is always a metatype corresponding to the contextual
type of the type it's referencing. For some reason, the instance type
was also stored in this TypeLoc at random points in semantic analysis.
Under the assumption that this instance type is always going to be the
instance type of the contextual type of the expression, introduce
a number of simplifications:
1) Explicit TypeExpr nodes must be created with a TypeRepr node
2) Implicit TypeExpr nodes must be created with a contextual type
3) The typing rules for implicit TypeExpr simply opens this type
wrapped value placeholder in an init(wrappedValue:) call that was previously
injected as an OpaqueValueExpr. This commit also restores the old design of
OpaqueValueExpr.
Add the `@differentiable` function conversion pipeline:
- New expressions that convert between `@differentiable`,
`@differentiable(linear)`, and non-`@differentiable` functions:
- `DifferentiableFunction`
- `LinearFunction`
- `DifferentiableFunctionExtractOriginal`
- `LinearFunctionExtractOriginal`
- `LinearToDifferentiableFunction`
- All the AST handling (e.g. printing) necessary for those expressions.
- SILGen for those expressions.
- CSApply code that inserts these expressions to implicitly convert between
the various function types.
- Sema tests for the implicit conversions.
- SILGen tests for the SILGen of these expressions.
Resolves TF-833.
