The generated header did not compile due to a bug that prevented us from
referencing the correct namespaces derived from the nominal type's name
(an extension does not have a name). Moreover, we did not generate
forward declarations for the members of the extensions for classes and
enums (but we did for structs). This change also removes a workaround
that emitted String::Index as _String_Index.
rdar://153221450
There are two main scenarios when printing a compatibility header that
references a @cdecl enum defined in Swift code. (1) When defined in the
same module as it's used we can print the definition normally and then
reference it. (2) When used in a different mode we need to print a
forward declaration before we can reference it.
This change adds printing the forward declaration and fix an issue where
the compiler would instead print an @include of the Swift module. The
import of the Swift module would work only in a local scenario where a
compatibility header and module would be generated under the same name.
However for a distributed frameworks we do not distribute the
compatibility header so this strategy doesn't work. Relying on a forward
declaration should be more reliable in all cases but clients may need to
import the other compatibility header explicitly.
Print @cdecl enums in the C section of the compatibility header. Use and
extend the macros to support C compiler clients.
The macro is adapted to the features supported by the client compiler.
It uses an Objective-C style macro with raw type when available and
fallbacks to a simple typedef for C compatibility.
Add a block for C clients in the compatibility header. This block
contains only the `@cdecl` functions that are printed using only C
types.
This C block is printed above the Objective-C and C++ blocks as if we
add support for `@cdecl` types other languages should be able to
reference them in function signatures. Other languages block don't
duplicate printing the `@cdecl` functions either as they are already
accessible to them.
Print the type traits in reverse interop to enable the use of foreign
reference type in generics like Swift arrays. Also make sure optional
foreign reference types can be passed around as raw pointers.
rdar://108139769
Compare the names of all extension members first, before attempting weirder and more expensive comparisons like stringified type and mangled name. This gives us a sort order that’s a little more comprehensible to humans.
Factor ModuleContentsWriter’s declaration sorting out into a separate helper function, and additionally rework that function into a series of abstract comparisons supported by various helper functions. This declutters the function and makes the high-level logic it implements much more clear, at the cost of hiding much of the control flow inside a macro.
This also makes a very small change to the handling of generic signature comparisons: declarations without a generic signature will be factored into comparisons by comparing an empty string.
PrintAsClang is supposed to emit declarations in the same order regardless of the compiler’s internal state, but we have repeatedly found that our current criteria are inadequate, resulting in non-functionality-affecting changes to generated header content. Add a diagnostic that’s emitted when this happens soliciting a bug report.
Since there *should* be no cases where the compiler fails to order declarations, this diagnostic is never actually emitted. Instead, we test this change by enabling `-verify` on nearly all PrintAsClang tests to make sure they are unaffected.
This did demonstrate a missing criterion that only mattered in C++ mode: extensions that varied only in their generic signature were not sorted stably. Add a sort criterion for this.
A couple of the rules that `ModuleContentsWriter::write()` uses to sort declarations didn’t actually work because of an incorrect predicate. In addition, there were a number of situations that could come up in C++ interop (where overloading is permitted) where extensions could not be sorted. Rework extension sorting to look for more kinds of differences between extension members.
Eliminates extraneous newlines between top-level Objective-C declarations in `-emit-objc-header` headers. Specifically, there should now always be exactly one—no more, no less—empty line between `@end` and whatever follows it.
Besides being more aesthetically pleasing, this eliminates ordering-dependent behavior where PrintAsClang would print an extra newline when visiting an empty extension, which meant that the order in which empty and non-empty extensions were visited during printing could result in whitespace differences in the compiler output. Printing the blank line is now conditional on whether `tell()` indicates that characters were actually written to the output.
Fixes rdar://143533893.
This change addresses the following issue: when an error is being wrapped in a warning, the diagnostic message will use the wrapper's `DiagGroupID` as the warning's name. However, we want to retain the original error's group for use. For example, in Swift 5, async_unavailable_decl is wrapped in error_in_future_swift_version. When we print a diagnostic of this kind, we want to keep the `DiagGroupID` of `async_unavailable_decl`, not that of `error_in_future_swift_version`.
To achieve this, we add `DiagGroupID` to the `Diagnostic` class. When an active diagnostic is wrapped in DiagnosticEngine, we retain the original `DiagGroupID`.
For illustration purposes, this change also introduces a new group: `DeclarationUnavailableFromAsynchronousContext`.
With this change, we produce errors and warnings of this kind with messages like the following:
```
global function 'fNoAsync' is unavailable from asynchronous contexts [DeclarationUnavailableFromAsynchronousContext]
global function 'fNoAsync' is unavailable from asynchronous contexts; this is an error in the Swift 6 language mode [DeclarationUnavailableFromAsynchronousContext]
```
Currently, we do not support exporting zero-sized value types from Swift
to C++. It needs some work on our end as these types are not part of the
lowered signature. In the meantime, this PR makes sure that common (but
not all) zero sized types are properly marked as unavailable. This is
important as the proper diagnostic will give users a hint how to work
around this problem. Moreover, it is really easy to hit this when
someone is experimenting with interop, so it is important to not have a
cryptic failure mode.
rdar://138122545
Some macro-generated declarations are not being printed in the
Obj-C/C++ generated header. Members introduced by attached `member`
macros on a type appear to be fine, but those introduced by a
attached `peer` or freestanding `declaration` macros don't show up.
This change updates the header writer to call `getAllMembers`
throughout instead of `getMembers`, which makes sure that everything
gets collected. Likewise, we update the top-level logic from
`getTopLevelDecls` to `getTopLevelDeclsWithAuxiliaryDecls` to pick
up freestanding decls introduced at file scope.
Fixes https://github.com/swiftlang/swift/issues/68170.
It is really involved to change how methods and classes are emitted into
the header so this patch introduces the impression of nested structs
through using statements and still emits the structs themselves as top
level structs. It emits them in their own namespace to avoid name
collisions. This patch also had to change some names to be fully
qualified to avoid some name lookup errors in case of nested structs.
Moreover, nesting level of 3 and above requires C++17 because it relies
on nested namespaces. Only nested structs are supported, not nested
classes.
Since this patch is already started to grow quite big, I decided to put
it out for reviews and plan to address some of the shortcomings in a
follow-up PR.
rdar://118793469
Some fields in the AST are cached values that are populated lazily. We
should not use those values directly as in case they are not yet
computed we get back null pointers. Use ASTContext instead which can
call the slow path if the cache is not yet populated.
rdar://132746445
The code already forward declared strutcs and enums. This patch extends
the logic to also forward declare classes. Unfortunately, there was some
fallout because some traits ended up defined multiple times for some
classes, so the code is now extended to only conditionally emit these
traits if no forward declaration was emitted for the type yet.
rdar://124022242
Because the underlying API for fetching top-level decls returns them in an unspecified order, PrintAsClang sorts the decls before printing them to make the output order more stable. However, the rules currently implemented have at least one known defect (they compare only the unqualified name of a nested class, so two nested classes with the same Swift name sort in an arbitrary order), and there are likely many more.
Add a fallback rule which sorts declarations by their mangled name; this should at least distinguish all non-colliding ValueDecls from each other, albeit according to fairly opaque criteria. Additionally add a rule to help distinguish extensions with very similar content, and tweak other logic so that the comparison function is less likely to give up early rather than continuing to look for a usable difference.
Fixes rdar://129485103.
The generated thunks for functions can refer to some internal methods of
their arguments. As a result, those generated thunks should always be
after the definitions of the corresponding argument types. The printer
ordered the declarations by name, and Swift had the convention starting
types with upper case letters and functions with lower case letters.
This naming convention together with the ordering resulted in the
correct ordering in most of the cases. There were a couple of exceptions
when people diverged from the naming conventions or wanted to export
operators. This patch fixes this problem by always ordering type decls
before function decls.
rdar://129276354
In some cases, the reverse interop generated both a forward declaration and a
definition with unavailable attribute in the C++ header. Unfortunately, the
kinds of these symbol did not match. The forward declaration was templated
while the definition was not. The forward declaration has the correct kind,
so this patch extends the printing of unavailable definitions to include the
generic arguments.
rdar://119835933
Although I don't plan to bring over new assertions wholesale
into the current qualification branch, it's entirely possible
that various minor changes in main will use the new assertions;
having this basic support in the release branch will simplify that.
(This is why I'm adding the includes as a separate pass from
rewriting the individual assertions)
As we do with Swift structs, emit the members of extensions of Swift
enums into the corresponding C++ class. This includes exposing more of
the Optional API from the standard library into Swift.
The checking should be as simple as checking for `isAny()` or `isAnyObject()`
on an existential type because all of the non-inverse protocol requirements
are going to form a protocol composition type.
[interop][SwiftToCxx] Ignore delayedMembers in C++
This fixes a bug where an ObjC @interface declaration is emitted for a
class that has a member that isn't emitted.
Resolves rdar://119835836
The frontend option '-clang-header-expose-module' allows the user to specify that APIs from an imported module have been exposed in another generated header, and thus APIs that depend on them can be safely exposed in the current generated header.
Each emitted declaration is annotated with the external_source_symbol with its own USR, to allow Clang's indexer to recognize this declaration as a Swift declaration with a specific USR
This is done by exposing nested Index and UTF8View type in String. Nested types aren't
supported yet, so we need to employ a number of hacks to emit them.
