While returning the actual child vector from
`ExpandChildTypeRefinementContextsRequest` is a nice idea, it is both
inefficient (the vector gets copied in and out) and kind of inaccurate, since
the vector remains mutable after the initial expansion and may gain additional
children as macros are lazily expanded.
Also rename it to `getExplicitReturnStmts` for clarity and have it
take a `SmallVector` out parameter instead as a small optimization and
to discourage use of this new method as an alternative to
`AnyFunctionRef::bodyHasExplicitReturnStmt`.
If there is only one connected component, we don't actually have
to form the array of components at all. Re-organize the code so
that we can short-circuit this decision as soon as possible.
Changing the members of these structs are still API and ABI breaking
changes but they are not as fragile as C++ classes where adding/removing
virtual functions or doing other changes can also result in breaking the
ABI.
rdar://119319825
Tuples in general do have conversions but an empty tuple or `Void`
doesn't, which means that if a type variable has a subtype binding
to `Void` it should be safe to prioritize.
Currently, C++ types cannot appear in resilient interfaces. There are
some cases where this is overly restrictive. We plan to improve the
logic to detect what types should not appear on resilient moduel
boundaries. In the meantime, this PR introduces a flag to disable these
errors. Users relying on this flag are on their own, this should only be
a temporary workaround until we land further improvements to this
diagnostic.
rdar://137457118
The Clang importer maps arbitrary attributes spelled with `swift_attr("...")`
over to Swift attributes, using the Swift parser to process those attributes.
Extend this mechanism to allow `swift_attr` to refer to an attached macro,
expanding that macro as needed.
When a macro is applied to an imported declaration, that declaration is
pretty-printed (from the C++ AST) to provide to the macro implementation.
There are a few games we need to place to resolve the macro, and a few more
to lazily perform pretty-printing and adjust source locations to get the
right information to the macro, but this demonstrates that we could
take this path.
As an example, we use this mechanism to add an `async` version of a C
function that delivers its result via completion handler, using the
`@AddAsync` example macro implementation from the swift-syntax
repository.
My change to preserve type sugar in type transforms introduced a
performance regression here, because openType() is called for
every disjunction choice.
My hope is to optimize the TypeAliasType representation and remove
this at some point, but for now, let's just restore the old
desugaring behavior in this case.
