KeyPath's getter/setter/hash/equals functions have their own calling
convention, which receives generic arguments and embedded indices from a
given KeyPath argument buffer.
The convention was previously implemented by:
1. Accepting an argument buffer as an UnsafeRawPointer and casting it to
indices tuple pointer in SIL.
2. Bind generic arguments info from the given argument buffer while emitting
prologue in IRGen by creating a new forwarding thunk.
This 2-phase lowering approach was not ideal, as it blocked KeyPath
projection optimization [^1], and also required having a target arch
specific signature lowering logic in SIL-level [^2].
This patch centralizes the KeyPath accessor calling convention logic to
IRGen, by introducing `@convention(keypath_accessor_XXX)` convention in
SIL and lowering it in IRGen. This change unblocks the KeyPath projection
optimization while capturing subscript indices, and also makes it easier
to support WebAssembly target.
[^1]: https://github.com/apple/swift/pull/28799
[^2]: https://forums.swift.org/t/wasm-support/16087/21
An initial implementation of a rework in how
we prevent noncopyable types from being
substituted in places they are not permitted.
Instead of generating a constraint for every
generic parameter in the solver, we produce
real Copyable conformance requirements. This
is much better for our longer-term goal of
supporting `~Copyable` in more places.
Previously, fully qualified types would be missing for global vars and
properties in `.swiftinterface` files that were emitted lazily.
Adding the test case also revealed that PatternBindingDecls needed to be
typechecked before lazy module serialization as well.
Previously, `usesFeatureRethrowsProtocol()` could mistakenly return false in
lazy typechecking mode since the types in an inheritance clause might not have
been resolved yet.
Previously, a `.swiftinterface` emitted in lazy typechecking mode would fall
back on printing the `TypeRepr` of entries in the inheritance clause of a
declaration since inherited types could be unresolved.
Wrap the `InheritedEntry` array available on both `ExtensionDecl` and
`TypeDecl` in a new `InheritedTypes` class. This class will provide shared
conveniences for working with inherited type clauses. NFC.
These allow multi-statement `if`/`switch` expression
branches that can produce a value at the end by
saying `then <expr>`. This is gated behind
`-enable-experimental-feature ThenStatements`
pending evolution discussion.
Don't print the uses of freestanding declaration macros in Swift
interfaces. Instead, print the declarations they produce, as
appropriate.
Fixes rdar://113730928.
Type-checking patterns is not fully requestified, so the
ASTPrinter should not assume this has been filled in.
It appears sufficient to just visit the associated VarDecl,
as we do already.
This logic may override the module considered as the source of a decl to
ensure that only visible modules are printed in the swiftinterface. This
change updates it to consider @_spiOnly imported modules from the local
module as visible when printing the private swiftinterface only.
rdar://103325332
This attribute can be attached to a noncopyable struct to specify that its
storage is raw, meaning the type definition is (with some limitations)
able to do as it pleases with the storage. This provides a basis for
implementing types for things like atomics, locks, and data structures
that use inline storage to store conditionally-initialized values.
The example in `test/Prototypes/UnfairLock.swift` demonstrates the use
of a raw layout type to wrap Darwin's `os_unfair_lock` APIs, allowing
a lock value to be stored inside of classes or other types without
needing a separate allocation, and using the borrow model to enforce
safe access to lock-guarded storage.
* Add experimental feature `PlaygroundExtendedCallbacks` which passes more information in `-playground` callbacks
Adds the experimental feature `PlaygroundExtendedCallbacks` which (when `-playground` is also passed) causes the playground transform to use alternate forms of the result-value, scope-entry, and scope-exit callbacks that include the module name and file path of the source file.
The previous callbacks included integers for the module number and file number, but this was cumbersome to use because it required the caller to create source symbols with magical names formed from the module name and file path that the playground transform knew how to look up.
The extended callbacks in the experimental feature instead pass these strings as string literals. This is an experimental feature because of the need to measure the performance impact, and because of the need to provide an option to control which set of callbacks to use so that existing clients of the playground transform can opt into it when ready. It's also likely that we'll want to pass more information in the extended callbacks, such as an indication of the kind of result is being logged (e.g. a loop iteration variable vs a return statement vs a variable assignment). So this should be considered the first of a series of experimental improvements that will then be pitched as an actual, non-experimental v2.0 of the playground transform callback API. Because of the nature of how the playground transform is used, it's much easier to iterate on the functionality in the form of an experimental feature rather than using only desktop debug builds of the Swift compiler.
Changes:
- define a new experimental feature called `PlaygroundExtendedCallbacks`
- modify the playground transform step in sema to pass the module name and file name literals when the experimental feature is set
- add a unit test for the extended callbacks
There is no change in behaviour when `PlaygroundExtendedCallbacks` is not enabled.
rdar://109911742
Co-authored-by: Brent Shank <bshank@apple.com>
This basically undoes 3da6fe9c0d, which in hindsight was wrong.
There were no other usages of TypeArrayView anywhere else except for
GenericSignature::getGenericParams(), and it was almost never what
you want, so callers had to convert back and forth to an ArrayRef.
Remove it.
Upcoming and experimental features are supported via command-line flags
and also in the SwiftPM manifest. Introduce it as an experimental
feature so that it can be enabled via SwiftPM without having to resort
to unsafe flags.
The `StrictConcurrency` experimental feature can also provide a
strictness level in the same manner as `-strict-concurrency`, e.g.,
`StrictConcurrency=targeted`. If the level is not provided, it'll be
`complete`.
Note that we do not introduce this as an "upcoming" feature, because
upcoming features should be in their final "Swift 6" form before
becoming available. We are still tuning the checking for concurrency.
The reason why I am doing this is that this was not part of the original
evolution proposal (it was called an extension) and after some discussion it was
realized that partial consumption would benefit from discussion on the forums.
rdar://111353459
This is phase-1 of switching from llvm::Optional to std::optional in the
next rebranch. llvm::Optional was removed from upstream LLVM, so we need
to migrate off rather soon. On Darwin, std::optional, and llvm::Optional
have the same layout, so we don't need to be as concerned about ABI
beyond the name mangling. `llvm::Optional` is only returned from one
function in
```
getStandardTypeSubst(StringRef TypeName,
bool allowConcurrencyManglings);
```
It's the return value, so it should not impact the mangling of the
function, and the layout is the same as `std::optional`, so it should be
mostly okay. This function doesn't appear to have users, and the ABI was
already broken 2 years ago for concurrency and no one seemed to notice
so this should be "okay".
I'm doing the migration incrementally so that folks working on main can
cherry-pick back to the release/5.9 branch. Once 5.9 is done and locked
away, then we can go through and finish the replacement. Since `None`
and `Optional` show up in contexts where they are not `llvm::None` and
`llvm::Optional`, I'm preparing the work now by going through and
removing the namespace unwrapping and making the `llvm` namespace
explicit. This should make it fairly mechanical to go through and
replace llvm::Optional with std::optional, and llvm::None with
std::nullopt. It's also a change that can be brought onto the
release/5.9 with minimal impact. This should be an NFC change.
