In theory, we shouldn't need to deserialize @_implementationOnly dependencies. However,
potential decl recovery issues may bring down lldb if we insist on not importing these
dependencies, resulting in bad user experience as a result. This patch adds an internal
option to allow importing them and it should only be set by lldb and other tools.
rdar://65570721
Add `async` to the type system. `async` can be written as part of a
function type or function declaration, following the parameter list, e.g.,
func doSomeWork() async { ... }
`async` functions are distinct from non-`async` functions and there
are no conversions amongst them. At present, `async` functions do not
*do* anything, but this commit fully supports them as a distinct kind
of function throughout:
* Parsing of `async`
* AST representation of `async` in declarations and types
* Syntactic type representation of `async`
* (De-/re-)mangling of function types involving 'async'
* Runtime type representation and reconstruction of function types
involving `async`.
* Dynamic casting restrictions for `async` function types
* (De-)serialization of `async` function types
* Disabling overriding, witness matching, and conversions with
differing `async`
We have landed support for serialization Clang function types, but there is
still work to be done here.
Moreover, we should use a consistent style with `[(FIXME|TODO|NOTE): label]`,
instead of using different styles throughout the code.
Introduce 'TypeCheckSingleASTNode' mode that only type checks single body
element and dependent necessities (i.e. referencing ValueDecls and their
dependencies).
Renamed swift::typeCheckAbstractFunctionBodyAtLoc() to
swift::typeCheckASTNodeAtLoc(DeclContext *, SourceLoc). That type checks
innermost 'ASTNode' at the location. Also, 'TypeCheckSingleASTNode' mode
skips type checking any "body" of the node (i.e. BraceStmt elements for
function body, if statement body, closure body, etc.)
Added on-demand type checking using it:
- VarDecl in TapExpr
- ParamDecl in ClosureExpr
- Return type of ClosureExpr
- Binding value in control statements
(e.g. ForEachStmt, SwitchStmt, DoCatchStmt, etc.)
rdar://problem/63932852
Introsuce a new "forward" algorithm for trailing closures where
the unlabeled trailing closure argument matches the next parameter in
the parameter list that can accept an unlabeled trailing closure.
The "can accept an unlabeled trailing closure" criteria looks at the
parameter itself. The parameter accepts an unlabeled trailing closure
if all of the following are true:
* The parameter is not 'inout'
* The adjusted type of the parameter (defined below) is a function type
The adjusted type of the parameter is the parameter's type as
declared, after performing two adjustments:
* If the parameter is an @autoclosure, use the result type of the
parameter's declared (function) type, before performing the second
adjustment.
* Remove all outer "optional" types.
For example, the following function illustrates both adjustments to
determine that the parameter "body" accepts an unlabeled trailing
closure:
func doSomething(body: @autoclosure () -> (((Int) -> String)?))
This is a source-breaking change. However, there is a "fuzzy" matching
rule that that addresses the source break we've observed in practice,
where a defaulted closure parameter precedes a non-defaulted closure
parameter:
func doSomethingElse(
onError: ((Error) -> Void)? = nil,
onCompletion: (Int) -> Void
) { }
doSomethingElse { x in
print(x)
}
With the existing "backward" scan rule, the trailing closure matches
onCompletion, and onError is given the default of "nil". With the
forward scanning rule, the trailing closure matches onError, and there
is no "onCompletion" argument, so the call fails.
The fuzzy matching rule proceeds as follows:
* if the call has a single, unlabeled trailing closure argument, and
* the parameter that would match the unlabeled trailing closure
argument has a default, and
* there are parameters *after* that parameter that require an argument
(i.e., they are not variadic and do not have a default argument)
then the forward scan skips this parameter and considers the next
parameter that could accept the unlabeled trailing closure.
Note that APIs like doSomethingElse(onError:onCompletion:) above
should probably be reworked to put the defaulted parameters at the
end, which works better with the forward scan and with multiple
trailing closures:
func doSomethingElseBetter(
onCompletion: (Int) -> Void,
onError: ((Error) -> Void)? = nil
) { }
doSomethingElseBetter { x in
print(x)
}
doSomethingElseBetter { x in
print(x)
} onError: { error in
throw error
}
Introduce a new frontend flag -enable-volatile-modules to trigger
loading swiftmodule files as volatile and avoid using mmap. Revert the
default behavior to using mmap.
This reverts commit 621b3b4223.
The driver is double faulting on my Linux box (Fedora 32 / x86-64). It
crashes due to heap corruption, then hangs trying to introspect and
print the stack. There also appears to be an unrelated(?) uninitialized
memory error that valgrind detects (as opposed to malloc's own self
diagnostics).
This makes it easier to specify OptionSet arguments.
Also modify appropriate uses of ModuleDecl::ImportFilter to take
advantage of the new constructor.
Since libDemangling is included in the Swift standard library,
ODR violations can occur on platforms that allow statically
linking stdlib if Swift code is linked with other compiler
libraries that also transitively pull in libDemangling, and if
the stdlib version and compiler version do not match exactly
(even down to commit drift between releases). This lets the
runtime conditionally segregate its copies of the libDemangling
symbols from those in the compiler using an inline namespace
without affecting usage throughout source.
to match the one specified in LLVM's Mach-O parser.
Otherwise LLDB could not possibly find it!
This name is used by the swift -modulewrap subcommand, which is currently unused
on Darwin, and primarily intended for use under Linux. However, it may be useful
to better support static archives (.a) files with Swift debug info in the
future. To fully support this, dsymutil and LLDB need to know to look for Swift
AST sections in Mach-O objects other than .dSYM bundled.
Implementation note: It would be nice to get the section name out of libObject,
but with the current architecture this needs a major refactoring that didn't
seem justified, given that there is an end-to-end test to prevent such a mishap
in the future.
<rdar://problem/63991514>
-enable-experimental-private-intransitive-dependencies -> -enable-direct-intramodule-dependencies
-disable-experimental-private-intransitive-dependencies -> -disable-direct-intramodule-dependencies
While we're here, rename DependencyCollector::Mode's constants and clean
up the documentation.
"Private Intransitive Dependencies" differ from the status quo by no
longer requiring the concept of a "cascading dependency edge". This is
because the request evaluator automatically tracks, records, and replays
the names looked up while a given file is being processed by the
frontend. To remove transitivity from the swiftdeps files, each
primary file processed by the Swift frontend is charged for *all* name
lookups that occur while it is being processed. Further, because of
the replay step, lookups hidden behind cached requests are now entirely
visible to the dependency tracking code.
The net result is that all formerly implicit transitivity in the
dependency graph has been made completely explicit and direct. This
establishes a tighter overall dependency structure for each individual
file, and results in a remarkable decrease in the amount of
files that are rebuilt for any particular change.
This feature can be disabled via
-disable-experimental-private-intransitive-dependencies, which will
cause a return to the cascading status quo.
