Most `SemaAnnotator`s don’t actually care about the char source range. Instead, they only care about the start location of the reference, which is also included in `SourceRange`. Computing a `CharSourceRange` from a `SourceRange` is kind of expensive because it needs to start a new lexer.
To avoid this overhead, pass `SourceRange` to `SemaAnnotator::passReference` and related functions and let the clients compute the `CharSourceRange` when needed.
This reduces the overhead of index-while-building by about 10%.
Extend the module trace format with a field indicating whether a given
module, or any module it depends on, was compiled with strict memory
safety enabled. This separate output from the compiler can be used as
part of an audit to determine what parts of Swift programs are built
with strict memory safety checking enabled.
CAS needs to determine if an output is needed very early (when computing supplementary outputs) in the pipeline so we can do caching and replays.
Resolves: rdar://141850408
Emitting fine-grained module tracing files during full compilation should be sufficient for the use case in design. Other
swift-frontend invocations, such as emitting modules, doesn't need fully type-checked AST to perform the action, therefore
requesting/walking the fully typechecked AST to emit those tracing files may introduce performance overhead.
When lazy type checking is enabled, the generated AST to walk on may contain function bodies that
are not fully type-checked. Walking on these function bodies could expose latent bugs.
Resolves rdar://140818215
CAS support in compiler relies on supplementary paths to decide the mapping between input and output files. Therefore, we
have to compute the paths of the module ObjC trace files in this canonical place to have CAS support for
this newly added ObjC message trace files.
To simplify post-processing of these collected ObjC message traces, this PR teaches the compiler to atomically append the content of
ObjC message trace to a file location provided by the build environment, instead of emitting multiple files.
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)
LLVM is presumably moving towards `std::string_view` -
`StringRef::startswith` is deprecated on tip. `SmallString::startswith`
was just renamed there (maybe with some small deprecation inbetween, but
if so, we've missed it).
The `SmallString::startswith` references were moved to
`.str().starts_with()`, rather than adding the `starts_with` on
`stable/20230725` as we only had a few of them. Open to switching that
over if anyone feels strongly though.
* Factor out ASTContext plugin loading to newly introduced 'PluginLoader'
* Insert 'DependencyTracker' to 'PluginLoader'
* Add dependencies right before loading the plugins
rdar://104938481
Calling getImportedModules requires to list the desired kind of imports.
With the new kind of imports this has become cumbersome. Let's simplify
it by offering common sets of imports. Advanced call sites can still
list the desired imports explicitly.
When loading plugins from `-plugin-path`, use the global `PluginRegistry` to keep a record of what's loaded. Emit these dependencies to the loaded module trace.
llvm-project `ErrorHandling.h` was updated to remove std::string. This
added a new `report_fatal_error` overload taking a `const Twine &`,
removed the overload that took `const std::string &`, and updated
`fatal_error_handler_t` to use `const char *` rather than `const
std::string &`.
Fix uses of these functions to take into account these updates. Note
that without the `const std::string &` overload, passing a `std::string`
into `report_fatal_error` now results in an ambiguous match between the
`StringRef` and `Twine` overloads so we need to be explicit about one or
the other.
Ensure that we only have one writer to the module trace file at a time by using LLVM's `LockFileManager` utilities, similarly to `ModuleInterfaceBuilder`
Resolves rdar://76743462
Follow programming guidelines for these getters more closely and have them return a non-owning view of the underlying data instead of relying on callers to take const references to the copy that is returned here.
T *const does not prevent logically non-const accesses to the underlying data, it merely indicates that the pointer value itself is const. This modifier can be cast off by a copy, so it's not generally what you want here. Switch to const T * instead.
