This affects module interfaces, interface generation in sourcekitd, and
diagnostics. Also fixes a fixit that was assuming the 'OSX' spelling when
computing the source range to replace.
Resolves rdar://problem/64667960
Most of the changes fall into a few categories:
* Replace explicit "x86_64" with %target-cpu in lit tests
* Cope with architecture differences in IR/asm/etc. macOS-specific tests
Previously we were linking in all SIL entities
if the input was a serialized non-SIB AST, and
`-disable-sil-linking` wasn't specified. However
none of the tests appear to want this behaviour.
Stop calling `SerializedSILLoader::getAll`, and
remove the `-disable-sil-linking` option, as this
is now the default behaviour.
The effective access of an overridden declaration is subject to
escalation by -enable-testing. When this is flag is enabled, an
interface containing an internal-overriding-public declaration will
still print `override`. This is because the effective access of the base
of the override is formally internal but effectively public.
Instead, use the formal access scope of the overridden declaration to
compute its access relative to the override. While I'm here, catch the
case where the base declaration is `@usableFromInline` and therefore
*will* be printed in the interface. We treat these declarations as
effectively public for the purpose of printing `override`.
Resolves rdar://64969741
Make `make-unreadable` python 3 compatible. `POINTER` is now in
`ctypes` rather than in `ctypes.wintypes`. Use the new location which
is compatible across python 2 and python 3.
Fix the search order for module interfaces to respect the documented
behavior:
1. Local cache
2. Next to the .swiftinterface
3. Prebuilt cache
Fixing the order of the last two allows the XcodeDefault compiler to
prioritize swiftmodule files with SPI information.
rdar://problem/64097455
Swift calls the architecture x86_64, OpenBSD calls it amd64. If we use
run_cpu in lit.cfg as-is, then we may need to duplicate lines in each
test for 'x86_64' and 'amd64', which puts a maintenance burden on unit
test developers to ensure they are duplicating changes to each line.
Instead, alias 'amd64' to 'x86_64' for `run_cpu`, but keep the platform
module path referring to 'amd64', in order to distinguish the target
architecture name and the Swift architecture name. This is particularly
relevant for the %target-.*-name pseudovariables used, which should
reference the Swift architecture names.
However, some unit tests are directly referencing %target-cpu directly,
which would break the aliasing. This is done only for swiftinterface
files, so a new substitution is defined in lit.cfg for these variables,
and the affected unit test cases are migrated.
For cases like this:
```swift
struct X {}
struct Y {}
func overloaded<T>(_ value: T) -> T { value }
func overloaded<T>(_ value: T) -> Int { 0 }
func test(x: inout X, y: Y) {
x = overloaded(y)
}
```
Solver would record a `IgnoreAssignmentDestinationType` fix per overload,
`diagnoseForAmbiguity` could be used to properly diagnose ambiguity cases
like that.
The `-force-single-frontend-invocation` flag predates WMO and is now an
alias for `-whole-module-optimization`. We should use the latter and let
the former fade into history.
This code should not yield the null Type() on failure. Instead, diagnose
the failure and yield an ErrorType. This matches what clients of this
function expect to signal failure.
Include a regression test that exercises a common failure mode if we
don't do this: module interface loading crashes.
rdar://62745419
Master source compat suite job #4912 failed to build ReactiveKit because Combine does not have an interface for older iOS architectures that were retired before it was released, and a change to module loading diagnosis started to consider switfinterfaces. Add a test case that covers this situation.
Before this change, we would emit the warning only for swiftmodules; however,
it may be the case that only a swiftinterface (of a different arch) is present
but no swiftmodule is present.
Fixes rdar://problem/50905075.
The standard library (and other Swift modules built by our CMake build system)
has been building module files with an architecture only (e.g., x86_64.swiftmodule)
rather than a proper module triple (x86_86-apple-macosx10.15,
x86_64-apple-ios13.0-simulator, etc.), unlike every other build
system. There are hacks in the compiler and other tools to cope with
this unnecessary build difference. Fix the module file names so we'll
be able to remove the hacks later.
Fixes rdar://problem/49071536.
When we are printing Swift interface, we have to skip the override keyword
if the overriden decl is invisible from the interface. Otherwise, an error
will occur while building the Swift module because the overriding decl
doesn't override anything.
We couldn't skip every `override` keywords because they change the
ABI if the overriden decl is also publicly visible.
For public-override-internal case, having `override` doesn't have ABI
implication. Thus we can skip them.
rdar://58562780
Protocol conformances can be implied by super classes. When collecting the list
of IncludedProtocols, we should traverse the class inheritance chain and collect
all protocols.
Without this change, these implied conformances will be printed again via a synthesized
extension at the end of the Swift interface file, leading to an error of redundant
conformances when building modules.
rdar://58563540
Currently, when a swiftinterface file fails to load, we emit the specific diagnostics for the failures, followed by a generic “failed to load module ‘Foo’” message. This PR improves that final diagnostic, particularly when the cause may be that the interface was emitted by a newer compiler using backwards-incompatible syntax.
The only initialization these class objects need is ObjC realization, which can be done
fast with `objc_opt_self` on recent Apple OSes. The cache check just adds code size and
dirties memory.
The test uses %target-run, which requires executing code. Android CI
cannot execute code because no device is attached. Marking the test as
executable skip it in the Android CI machines.
This test fails sometimes with something like:
SOURCE_DIR/test/ModuleInterface/ModuleCache/prebuilt-module-cache-forwarding.swift:47:8: error: malformed compiled module: BUILD_DIR/test-macosx-x86_64/ModuleInterface/ModuleCache/Output/prebuilt-module-cache-forwarding.swift.tmp/MCP/Lib-2XZQCTV44N470.swiftmodule
rdar://58578342
Manually expand out the few places where we had a `not ...` substitution
where the substituted value was invoking `env` to alter the environment.
This pattern is not portable and causes problems when using the
integrated shell, such as on Windows. This was identified during the
LLVM update.
This additionally corrects the use of `not` to indicate that it is
expecting a crashing failure. This was resulting in the Windows tests
failing as `not` was not expecting a crashing failure.
