Commit Graph

17564 Commits

Author SHA1 Message Date
Karoy Lorentey
26176cc929 [stdlib] Allow metatypes of noncopyable/nonescapable types to get compared 2025-03-24 12:16:08 -07:00
Karoy Lorentey
eba00922c3 [stdlib] Properly generalize ObjectIdentifier.init(Any.Type)
With Slava’s fix in https://github.com/swiftlang/swift/pull/79647, we can now refer to generalized type existentials in ABI stable code.
2025-03-24 12:16:07 -07:00
Karoy Lorentey
801385ad49 [stdlib] Generalize the signature of type(of:)
`func type(of:)` is wholly magical, but it does have a signature in the stdlib (mostly for documentation purposes), and it currently requires its input to be copyable and escapable.

`type(of:)` is actually usable on all entities, so it seems desirable to update its signature to reflect this.

Additionally, this seems like a good time to mark its exported symbol obsolete. I don’t expect anyone would ever link to it (unless there is/was a bug), so in theory we could also silently remove it — but explicitly marking it as legacy ABI seems the least risky option.
2025-03-24 12:16:06 -07:00
Karoy Lorentey
3c133d56c1 Revert "[stdlib] unsafeBitCast: generalize for ~Escapable input types"
This reverts commit 0d7cab589f1a7fbcb988b28aec50db768e018539.
2025-03-24 12:16:03 -07:00
Karoy Lorentey
8a7fcc3bbf [stdlib] Update syntax for conditional conformances to invertable protocols 2025-03-24 12:16:02 -07:00
Karoy Lorentey
538f2742ef [stdlib] Roll back generalization of the nil-coalescing operator ??
We cannot currently express its proper lifetime semantics: its result’s lifetime should depend on the intersection of the lifetime of the left argument and the lifetime of the result of the right argument.

`@lifetime(optional, defaultValue.result)` is what we want, but the `.result` part is not currently expressible. (Tying the dependency on the closure argument itself may be a viable option, but we aren’t confident enough to ship it like that yet.)
2025-03-24 12:16:01 -07:00
Karoy Lorentey
96fd749ecc [stdlib] Generalize unsafe buffer pointer operations on Slice
`Slice` continues to require `Element` to be copyable, so this is limited to`moveInitializeMemory`, `bindMemory`, `withMemoryRebound` and `assumingMemoryBound`. Generalizing these restores parity with the corresponding operations on unsliced buffer pointers, and reduces the need to artificially rebase the slices only to call these.
2025-03-24 12:16:01 -07:00
Karoy Lorentey
b7b75aeace [stdlib] Generalize Optional.unsafelyUnwrapped 2025-03-24 12:15:56 -07:00
Karoy Lorentey
cf35c65863 [stdlib] Generalize Unsafe[Mutable]BufferPointer.indices 2025-03-24 12:15:55 -07:00
Karoy Lorentey
0e55d96a22 [stdlib] Generalize Optional.take(), ?? and nil comparisons 2025-03-24 12:15:52 -07:00
Karoy Lorentey
84b63956bf [stdlib] Generalize Optional’s “some” initializer 2025-03-24 12:15:51 -07:00
Karoy Lorentey
463f35531c [stdlib] ObjectIdentifier: Work around issue with generalized existentials 2025-03-24 12:15:49 -07:00
Karoy Lorentey
4db7b7c69d [stdlib] Note some syntactic wrinkles 2025-03-24 12:15:47 -07:00
Karoy Lorentey
a52c060924 [stdlib] dependsOn → @lifetime 2025-03-24 12:15:46 -07:00
Karoy Lorentey
b9d7206028 [stdlib] Start generalizing Result
- Allow nonescapable `Success` types
- Generalize `Sendable` conformance
- Generalize `Result.mapError`
- Generalize `Result.get()`

Do not generalize `Result.init(catching:)` or higher-order functions involving the success value: they cannot possibly work correctly yet.
2025-03-24 12:15:44 -07:00
Karoy Lorentey
0d926834c8 [stdlib] ObjectIdentifier: Also allow nonescapable types 2025-03-24 12:15:44 -07:00
Karoy Lorentey
a0751a4a89 [stdlib] ObjectIdentifier: Allow initialization from metatypes of noncopyable types 2025-03-24 12:15:43 -07:00
Karoy Lorentey
fb3455eccf [stdlib] Generalize withExtendedLifetime; add extendLifetime 2025-03-24 12:15:42 -07:00
Karoy Lorentey
201430522b [stdlib] Generalize ManagedBufferPointer.== 2025-03-24 12:15:41 -07:00
Karoy Lorentey
351adc7d4a [stdlib] Enable NonescapableTypes 2025-03-24 12:15:41 -07:00
Karoy Lorentey
696bbac938 [stdlib] unsafeBitCast: generalize for ~Escapable input types 2025-03-24 12:15:39 -07:00
Karoy Lorentey
9b322e3e8e [stdlib] Optional: Allow non-escapable wrapped types 2025-03-24 12:15:38 -07:00
Karoy Lorentey
9d86a45af9 [stdlib] ExpressibleByNilLiteral: allow ~Escapable conformers 2025-03-24 12:15:37 -07:00
Karoy Lorentey
db92d2b915 [stdlib] Generalize MemoryLayout for ~Escapable types 2025-03-24 12:15:33 -07:00
Tony Allevato
815c3e8a6e Merge pull request #80238 from alexfh/patch-1
Fix a compilation error with LLVM 21
2025-03-24 14:54:51 -04:00
Mike Ash
6a383b270e [Runtime] Ignore some global constructor warnings.
We have a few constructor functions that aren't wrapped in SWIFT_ALLOWED_RUNTIME_GLOBAL_CTOR_BEGIN/SWIFT_ALLOWED_RUNTIME_GLOBAL_CTOR_END and which have started to produce warnings in a new clang version. Explicitly allow these constructors by adding those.

rdar://147703947
2025-03-24 12:43:03 -04:00
Guillaume Lessard
c21de062bb Merge pull request #78561 from glessard/rdar137710901-span-properties-prototype
[SE-0456] Span properties (part 1)
2025-03-24 08:47:47 -07:00
Holly Borla
498d7ccc17 Merge pull request #80187 from hborla/task-macro-improvements
[Macros] Update the name and argument list for the `@Task` function body macro.
2025-03-24 08:46:32 -07:00
Alastair Houghton
b032d704ed Merge pull request #80214 from al45tair/rename-unsafefirst
[Backtracing] Rename `unsafeFirst` to `consumingFirst`.
2025-03-24 14:32:44 +00:00
Alexander Kornienko
dd83d1e1e0 Fix a compilation error
This fixes a `copy constructor must pass its first argument by reference` compilation error when compiled with a recent enough Clang (after fe0d3e3764)
2025-03-24 11:46:10 +01:00
Hamish Knight
f8ab391737 Introduce type sugar for InlineArray (#80087)
* [CS] Decline to handle InlineArray in shrink

Previously we would try the contextual type `(<int>, <element>)`,
which is wrong. Given we want to eliminate shrink, let's just bail.

* [Sema] Sink `ValueMatchVisitor` into `applyUnboundGenericArguments`

Make sure it's called for sugar code paths too. Also let's just always
run it since it should be a pretty cheap check.

* [Sema] Diagnose passing integer to non-integer type parameter

This was previously missed, though would have been diagnosed later
as a requirement failure.

* [Parse] Split up `canParseType` 

While here, address the FIXME in `canParseTypeSimpleOrComposition`
and only check to see if we can parse a type-simple, including
`each`, `some`, and `any` for better recovery.

* Introduce type sugar for InlineArray

Parse e.g `[3 x Int]` as type sugar for InlineArray. Gated behind
an experimental feature flag for now.
2025-03-23 15:31:37 -07:00
Saleem Abdulrasool
9354fd91d1 Merge pull request #80200 from Steelskin/fabrice/restore-float-fenv-modules
Put float and stddef back inside the ucrt module
2025-03-22 20:36:57 -07:00
Alastair Houghton
8b15b05c63 Revert "[Concurrency] Provide a Swift interface for custom main and global executors." 2025-03-22 02:38:11 -07:00
Guillaume Lessard
7b035936e4 [temporary] disable spans over inline elements 2025-03-21 17:26:29 -07:00
Guillaume Lessard
d5cb7ddf64 [stdlib] changes to lifetime annotations
Changes to lifetime annotations implemented in https://github.com/swiftlang/swift/pull/80064
2025-03-21 17:24:45 -07:00
Alastair Houghton
75b9c8f2cf [Backtracing] Rename unsafeFirst to consumingFirst.
This seems like a better name.
2025-03-21 21:18:31 +00:00
Gábor Horváth
ebeee4d716 Merge pull request #80163 from swiftlang/gaborh/vector-span-property
[cxx-interop] Add span property to std::vector
2025-03-21 19:00:34 +00:00
Saleem Abdulrasool
90d9ce76ba Concurrency: handle android similar to Linux
Android should use the dispatch based, Linux platform executor. However,
Android reports as `os(Android)` rather than `os(Linux)`. This adjusts
the condition to ensure that we have a platform executor factory for
Android.
2025-03-21 11:00:08 -07:00
Fabrice de Gans
d5ea19405d Put float and stddef back inside the ucrt module
These were split out in #79751. However, this split is not needed for
these. Furthermore, modulemaps have bugs when it comes to re-exporting
some modules, resulting in missing exports.
2025-03-21 10:01:01 -07:00
Holly Borla
bac0a10ae2 [Macros] Update the name and argument list for the function body macro
that wraps a function or closure body in a new top-level task.
2025-03-21 06:21:45 -07:00
Alastair Houghton
8443b5f76c Merge pull request #79789 from al45tair/custom-executors
[Concurrency] Provide a Swift interface for custom main and global executors.
2025-03-21 09:05:03 +00:00
Dario Rexin
35a44e5168 [Concurrency] Fix templated code in Task+startSynchronously.swift (#80100)
* [Concurrency] Fix templated code in Task+startSynchronously.swift

rdar://147348183

The gyb code always dropped the `throws` modifier, no matter which case was being processed.

* Update Task+startSynchronously.swift.gyb

* Fix abi tests
2025-03-20 18:29:31 -07:00
Dave Lee
6ca8e0d502 [Concurrency] Change signposts to use decimal format for task ids (#80154)
Change task ids to be formatted using decimal instead of hex.

An example of previous signposts:

```
task=e resumefn=0x2807722bc jobPriority=0 isChildTask=0, isFuture=1 isGroupChildTask=0 isAsyncLetTask=0 parent=0 group=0x0 asyncLet=0x0
```

this will now show `task=14` instead of `task=e`
2025-03-20 13:23:37 -07:00
Gabor Horvath
3d6c714cd6 [cxx-interop] Add span property to std::vector
rdar://146944876
2025-03-20 16:15:06 +00:00
Mike Ash
1e265c24d7 Merge pull request #80096 from mikeash/task-group-cancellation-lock-fix
[Concurrency] Fix a race when using cancelAll on a task group concurrently with child tasks being removed.
2025-03-20 11:53:31 -04:00
Saleem Abdulrasool
43ec13a65b Merge pull request #80151 from Steelskin/fabrice/restore-ucrt-legacy-stdio-def
Restore legacy_stdio_definitions in ucrt.modulemap
2025-03-19 21:57:15 -07:00
Fabrice de Gans
7a965cee00 Restore legacy_stdio_definitions in ucrt.modulemap
This was removed in #79751 but is required in some cases.
2025-03-19 16:51:38 -07:00
Andrew Trick
955d089a90 CxxSpanReturnThunkBuilder: use _cxxOverrideLifetime(_:copying:)
Do not rely on the @_unsafeNonescapableResult attribute. That attribute is only
for temporarily working around bugs! And it only affects lifetime diagnostics within
the function. It has no affect on the caller's diagnostics, so it won't solve
this problem:

func macroGeneratedThunk() -> CxxSpan<Int> {
  return _unsafeRemoveLifetime(Span...)
}

We cannot simply add @_unsafeRemoveLifetime to the thunk, because SwiftSyntax
does not natively support the attribute. We don't want to add SwiftSyntax
support because this attribute will never be supported syntax!

Instead, use `_overrideLifetime` copying the `Void` type to remove a dependency:

func macroGeneratedThunk() -> CxxSpan<Int> {
  return _cxxOverrideLifetime(Span..., copying: ())
}
2025-03-19 11:59:05 -07:00
Andrew Trick
e91981266b Update CXXSpan for strict @lifetime 2025-03-19 11:59:04 -07:00
Andrew Trick
c6116be7b5 Update stdlib source with explicit @lifetime(copy self) 2025-03-19 11:59:04 -07:00