When an accessor macro adds a non-observing accessor to a property, it
subsumes the initializer. We had previously modeled this as removing
the initializer, but doing so means that the initializer could not be
used for type inference and was lost in the AST.
Explicitly mark the initializer as "subsumed" here, and be more
careful when querying the initializer to distinguish between "the
initializer that was written" and "the initializer that will execute"
in more places. This distinction already existed at the
pattern-binding level, but not at the variable-declaration level.
This is the proper fix for the circular reference issue described in
rdar://108565923 (test case in the prior commit).
Rather than requiring macro implementations to add required whitespace
and indentation, basic format all macro expansions. Right now this uses
the default four space indentation, we can consider having that inferred
later. Macros can opt-out of automatic formatting by implementing
`formatMode` and setting it to `.disabled`.
Also moves the extra newlines before/after expansions to a new "Inline
Macro" refactoring.
Resolves rdar://107731047.
The request returns the expanded buffer ID even if it failed to
typecheck the expanded buffer.
This makes refactoring 'Expand Macro' work regardless of the
typechecking results.
rdar://108530760
Expression macros ascribed to non-private contexts need private
discriminators so they don't conflict with other uses of the same macro
in other source files.
Thank you, Richard, for noticing this omission!
Add a CachedDiagnosticsProcessor that is a DiagConsumer can capture all
the diagnostics during a compilation, serialized them into CAS with a
format that can be replayed without re-compiling.
Eliminate another circular reference through macro expansion mangling
by adjusting the starting declaration context to ensure that it is from
a suitable "outer" context.
Fixes rdar://108511666.
Examine pack expansion pattern to determine whether expression
result could be discarded without a warning (applies to tuples
with a single unlabeled pack expansion element as well).
Resolves: rdar://108064941
If we have both loaded a swiftdoc, and the decl we
have should have had its doc comment serialized into
it, we can check it without needing to fall back
to the swiftsourceinfo.
This requires a couple of refactorings:
- Factoring out the `shouldIncludeDecl` logic
into `getDocCommentSerializationTargetFor` for
determining whether a doc comment should end up
in the swiftdoc or not.
- Factoring out `CommentProviderFinder` for searching
for the doc providing comment decl for brief
comments, in order to allow us to avoid querying
the raw comment when searching for it. This has the
added bonus of meaning we no longer need to fall
back to parsing the raw comment for the brief
comment if the comment is provided by another decl
in the swiftdoc.
This diff is best viewed without whitespace.
It doesn't seem like there's any client that's
actually taking advantage of setting it to `false`,
and its default value of `false` is more likely
than not going to cause clients to accidentally
miss comments that they may want. In fact, this
was exactly the case for code completion's brief
field. Finally, the parameter wasn't even
consistently applied, as we would attempt to
deserialize swiftdoc comments even if it were
`false`.
Code like that is usually indicative of programmer error, and does not
round-trip through module interface files since there is no source
syntax to refer to an outer generic parameter.
For source compatibility this is a warning, but becomes an error with
-swift-version 6.
Fixes rdar://problem/108385980 and https://github.com/apple/swift/issues/62767.
The reason why I am doing this is that currently SILGen knows when emitting said
closure that we are going to emit an error (that it does not have enough
information to emit itself since it doesn't know the caller), so doesn't emit
move checking markers. The result is that the move checker will not eliminate
any copies in the closure and thus will emit a "copy of noncopyable type found"
error and tell the user to file a bug. This just suppresses that.
rdar://108511866
This makes it so that the move address checker is not dependent on starting the
traversal at a base object. I also included verifier checks that the API can
visit all address uses for:
1. project_box.
2. alloc_stack.
3. ref_element_addr.
4. ref_tail_addr.
5. global_addr_inst.
this is because this visitor is now apart of the SIL API definition as being
able to enumerate /all/ addresses derived from a specific chosen address value.
This is a refactoring NFCI change.
rdar://108510644
* Factor out ASTContext plugin loading to newly introduced 'PluginLoader'
* Insert 'DependencyTracker' to 'PluginLoader'
* Add dependencies right before loading the plugins
rdar://104938481
The macro-resolution request for an attached macro was expressed in
terms of the custom attribute and the declaration context enclosing the
attribute. While the declaration context is the correct one for
resolving the types and arguments of the custom attribute, the
declaration provides a better anchor for cases where the same
attribute applies to multiple declarations, e.g., with
member-attribute macros, leading to false cyclic references.
We are likely to debugging more cyclic references involving macros in
the future, so improve the diagnostics to specify what operation we're
performing (e.g., macro resolution vs. expansion of a particular kind
of macro) and the macro name, when possible, so failures are easier to
diagnose. No functionality change for well-formed code.
Before this patch the parents of SILDebugScopes representing macro expansions
were missing the inlinedAt field, which resulted in incorrent LLVM IR being
produced. This is fixed by first computing the inlined call site for a macro
expansion and then computing the nested SILDebugScope for the ASTScope of the
expanded nodes; adding the inlinedAt field to all of levels of parent scopes.
rdar://108323748
