Global actors have been an accepted language feature for a couple of compiler releases now, but the feature definition was associated with the `--enable-experimental-concurrency` flag. This caused some `.swiftinterface`s containing global actor declarations to be unparsable because the logic for surrounding declarations with the `$GlobalActors` feature guard was incomplete (for example, classes with a global actor attribute were not guarded even though the declarations of global actor types were).
Rather than trying to fix the logic of `usesFeatureGlobalActors()`, enable it by default.
Adds a test that demonstrates that modules defining and using public global actors produce module interfaces that can be parsed successfully.
Resolves rdar://100150703
Controlled with a new flag '-direct-clang-cc1-module-build'
This will allow clients to formulate 'swift-frontend' invocations with fully-specified set of cc1 arguments (using '-Xcc -Xclang -Xcc <FLAG>') required for the PCM build, without having to go through the driver.
* [SILOptimizer] Add prespecialization for arbitray reference types
* Fix benchmark Package.swift
* Move SimpleArray to utils
* Fix multiple indirect result case
* Remove leftover code from previous attempt
* Fix test after rebase
* Move code to compute type replacements to SpecializedFunction
* Fix ownership when OSSA is enabled
* Fixes after rebase
* Changes after rebasing
* Add feature flag for layout pre-specialization
* Fix pre_specialize-macos.swift
* Add compiler flag to benchmark build
* Fix benchmark SwiftPM flags
Basic should not be allowed to link Parse, yet it was doing so
to allow Version to provide a constructor that would conveniently
parse a StringRef. This entrypoint also emitted diagnostics, so it
pulled in libAST.
Sink the version parser entrypoint down into Parse where it belongs
and point all the clients to the right place.
Remove the CMake configuration option `SWIFT_SWIFT_PARSER_ROUNDTRIP`.
Instead, whenever the "early" SwiftSyntax module is built, link the
Swift Swift parser into the compiler and related tools.
Introduce a new experimental feature `ParserRoundTrip` that can be
enabled to perform round-trip testing.
Having an out-of-line definition for the LangOptions constructor makes
it easier to enable experimental features one at a time for a build,
without rebuilding everything.
By using the keyword instead of the function, we actually get a much simpler
implementation since we avoid all of the machinery of SILGenApply. Given that we
are going down that path, I am removing the old builtin implementation since it
is dead code.
The reason why I am removing this now is that in a subsequent commit, I want to
move all of the ownership checking passes to run /before/ mandatory inlining. I
originally placed the passes after mandatory inlining since the function version
of the move keyword was transparent and needing to be inlined before we could
process it. Since we use the keyword now, that is no longer an issue.
In the backdeploy library, `ActiveTaskStatus` takes
`JobFlags::getPriority()` in the constexpr constructor; however,
`JobFlags::getPriority()` isn't constexpr, so this is incorrect.
Clang 5 (possibly earlier) has a bug where it does not merge default
template arguments properly for forwarded declarations. Include
`SmallVector.h` directly in this case. This will slightly increase
compile times, but only for very old compilers and only for files that
don't already include `SmallVector.h` (of which there's currently only
23).
Trivial conflict caused by the line above the
`IGM.constructInitialFnAttributes` change in `lib/IRGen/GenDecl.cpp`
having an extra argument passed in rebranch (due to the new LLVM API).
I am separating the concern of weakly linking symbols that are introduced at the deployment target from the concern of type checking this new type of potential unavailability.
Resolves rdar://97925900
On the BSDs when building with bootstrapping and therefore modules enabled,
this will result in a "declaration here is not visible" error. This is
_possibly_ because the assert implementation is not reexported from LLVM,
maybe due to libc differences. Regardless, this fixes the problem and
should be relatively inocuous to unconditionally include.
This was already enabled as part of `-enable-implicit-dynamic` but this
new flag allows turning on opaque type erasure all by itself whether or
not `dynamic` is added explicitly.
rdar://97375478
