When we are using diagnostic transactions to disable immediate emission
of diagnostics, `DiagnosticEngine::hadAnyError()` no longer accurately
reports whether an error occurred. Thread the DiagnosticTransaction
into the ConstraintSystem so we can also check whether it contains an
error before emitting the fallback diagnostic.
Fixes rdar://128272346.
This started out as a crash, where an expression macro could not be
defined in terms of one of the builtin macros (e.g., `#line`), because
we were expecting a macro expansion expression but didn't get one.
Easy fix.
However, this uncovered a second bug, which is that we couldn't handle
an expression macro expansino to `#line`. This is because we were
parsing the macro expansion buffer as "top level items", which treats
`#line` at the start of a line as a deprecated alias of
`#sourceLocation`. Switch over to parsing a single expression in these
contexts, and fix up an issue where `#isolation` didn't even have that
expression.
Fixes rdar://139372780.
The condition to make sure that the parser makes progress when there is
an ill-formed macro expansion to members was incorrect, causing an
infinite loop if there was at least one well-formed declaration
followed by ill-formed code (in this case, a `}`). Fix the condition.
Fixes rdar://137828917.
Find all the usages of `--enable-experimental-feature` or
`--enable-upcoming-feature` in the tests and replace some of the
`REQUIRES: asserts` to use `REQUIRES: swift-feature-Foo` instead, which
should correctly apply to depending on the asserts/noasserts mode of the
toolchain for each feature.
Remove some comments that talked about enabling asserts since they don't
apply anymore (but I might had miss some).
All this was done with an automated script, so some formatting weirdness
might happen, but I hope I fixed most of those.
There might be some tests that were `REQUIRES: asserts` that might run
in `noasserts` toolchains now. This will normally be because their
feature went from experimental to upcoming/base and the tests were not
updated.
Add flag `-load-resolved-plugin` to load macro plugin, which provides a
pre-resolved entry into PluginLoader so the plugins can be loaded based
on module name without searching the file system. The option is mainly
intended to be used by explicitly module build and the flag is supplied
by dependency scanner.
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.
Closures appearing in freestanding macro arguments don't have
discriminators assigned, since we don't actually emit them.
Similarly we skip recording opaque return types that appear in macro
arguments, since they don't get emitted.
However this logic didn't take delayed parsing into account, which must
save and restore the InFreestandingMacroArgument bit correctly.
As a result, if the freestanding macro argument contained a closure
which contained a local function with a declaration that has an
opaque return type, we would crash in serialization from attempting
to mangle an opaque return type nested inside of a closure without a
discriminator.
Fixes rdar://135445004
to verify ExportedSourceFileRequest == 0.
In release mode only non-zero stats are printed by default now.
Fix diagnostic when compiler is built without statistics support.
These currently end up as "open", which is incorrect (we need to make them
"final"). This cannot be addressed until we pass the lexical context
down to the macro, which is currently... a bit tricky.
The Clang importer maps arbitrary attributes spelled with `swift_attr("...")`
over to Swift attributes, using the Swift parser to process those attributes.
Extend this mechanism to allow `swift_attr` to refer to an attached macro,
expanding that macro as needed.
When a macro is applied to an imported declaration, that declaration is
pretty-printed (from the C++ AST) to provide to the macro implementation.
There are a few games we need to place to resolve the macro, and a few more
to lazily perform pretty-printing and adjust source locations to get the
right information to the macro, but this demonstrates that we could
take this path.
As an example, we use this mechanism to add an `async` version of a C
function that delivers its result via completion handler, using the
`@AddAsync` example macro implementation from the swift-syntax
repository.
My change to preserve type sugar in type transforms introduced a
performance regression here, because openType() is called for
every disjunction choice.
My hope is to optimize the TypeAliasType representation and remove
this at some point, but for now, let's just restore the old
desugaring behavior in this case.
In this PR i worked on replacing the word accessor in diagnostics with more user-facing terminologies like setter, getter, didSet Observer, and members based on the context of the message.
in some messages i didn't need to pass DescriptiveDeclKind instead i just changed the text copy itself.
i also updated tests, so you might find it easier to check my changes this way.
Please let me know if there's something i should've done in a better way, and request changes if needed. !
i can squash my commits after reviewing getting the PR Reviewed, just to make it easier to be checked commit by commit
Resolves#55887
Always add constraints, find fixes during simplify.
New separate fix for allow generic function specialization.
Improve parse heuristic for isGenericTypeDisambiguatingToken.
Degrade concrete type specialization fix to warning for macros.
When using `@DebugDescription`, only allow use of [LLDB Summary
Strings](https://lldb.llvm.org/use/variable.html#summary-strings) syntax from
`lldbDescription` properties. When `@DebugDescription` is applied to existing
`debugDescription` properties, escape any `$`, as the output of `debugDescription` is
never interpreted by LLDB.
Depends on #75305
From feedback, replace the name `_debugDescription`, which was confusing because of the
underscore, with `lldbDescription`. This new name also indicates that this property may
contain [LLDB Summary Strings](https://lldb.llvm.org/use/variable.html#summary-strings).
When building the ASTScope tree, we set the insertion location for the
macro expansion just before the end of the macro use. This is the right
approach for attached macros, because we need to put the entities
after the macro.
However, for freestanding macros, which can have trailing closures,
this location is a problem, because the scope for the macro
expansion ends up inside the scope for the trailing closure. Use the
start location of the freestanding macro instead.
Fixes rdar://130923190.
rdar://128870792
Synthesized files are treated as not having a parent source file, but that can cause issues for macro code. The declarations will have the source file of their use site declared as parent source file, which means they should be emitted into the same IGM. If we don't use the correct IGM, we can get inconsistencies when referencing the generated code, which will cause linking issues.
There is code in `IRGenDebugInfo.cpp` ([1]) that tries to move the
common part of the `filename` into the `directory`. If this test is
build with a temporary directory that shares a prefix with the current
working directory, the common part of the path will be moved into
`directory`, but since the test checks for an empty directory, it will
fail the test, even if the results are correct. This situation is not
uncommon in CI setups where each job might have an isolated temporary
directory sharing a prefix with the current working directory (example:
`/ci/build/XXX/tmp` and `/ci/build/XXX/cwd`).
Modify the regex to allow anything in the `directory` value, since it is
not really important for the test to prove that the debug information is
generated.`
[1]: 9c4232c89d/lib/IRGen/IRGenDebugInfo.cpp (L564-L585)
Separate swift-syntax libs for the compiler and for the library plugins.
Compiler communicates with library plugins using serialized messages
just like executable plugins.
* `lib/swift/host/compiler/lib_Compiler*.dylib`(`lib/CompilerSwiftSyntax`):
swift-syntax libraries for compiler. Library evolution is disabled.
* Compiler (`ASTGen` and `swiftIDEUtilsBridging`) only depends on
`lib/swift/host/compiler` libraries.
* `SwiftInProcPluginServer`: In-process plugin server shared library.
This has one `swift_inproc_plugins_handle_message` entry point that
receives a message and return the response.
* In the compiler
* Add `-in-process-plugin-server-path` front-end option, which specifies
the `SwiftInProcPluginServer` shared library path.
* Remove `LoadedLibraryPlugin`, because all library plugins are managed
by `SwiftInProcPluginServer`
* Introduce abstract `CompilerPlugin` class that has 2 subclasses:
* `LoadedExecutablePlugin` existing class that represents an
executable plugin
* `InProcessPlugins` wraps `dlopen`ed `SwiftInProcPluginServer`
* Unified the code path in `TypeCheckMacros.cpp` and `ASTGen`, the
difference between executable plugins and library plugins are now
abstracted by `CompilerPlugin`
Make sure we walk macro expansion decls to handle
cases where e.g a binding introduces control flow.
This should be a pretty uncommon case since bindings
introduced by macros aren't actually usable when
expanded in local contexts, but handle it all the
same.
Ignore any regions recorded while inside a macro
expansion, but account for any control flow that
may have happened such that the exit count is
correctly adjusted. This allows e.g throwing function
calls to behave correctly within the expansion.
rdar://129081384
