Cache the result of turning a `ValueDecl` into an `AvailabilityDomain`. Use
split caching to make the common case of the decl not representing an
availability domain efficient.
NFC.
Get rid of the boolean arguments for unavailability in AvailabilityQuery's
constructors and introduce a `asUnavailable()` modifier that can be used
instead in the contexts where unavailability is relevant.
Conditionally available opaque return types should support availability
conditions that are evaluated in any availability domain. Update
`ConditionallyAvailableSubstitutions` to model its conditions with
`AvailabilityQuery` instead of assuming that conditions are always a single
version query for the current platform.
Always give up early when attempting to deserialize a protocol
conformance broken by a context change. Don't attempt to replace missing
members of the conformance signature with invalid one, just mark the
whole protocol conformance as invalid.
The previous recovery logic, only for SourceKit mode and LLDB, was
inserting invalid conformances in the signature instead of dropping the
whole protocol conformance. It lead to failures later in the same
`finishNormalConformance` when accessing the invalid conformances.
rdar://98925842
While deserializing AST function types FunctionType and GenericFunctionType which include
lifetime dependencies and an implicit self parameter, we don't correctly populate
ASTExtInfoBuilder.lifetimeDependencies. We end up reading one dependency less due to
incorrect index calculation.
Unlike SILFunctionType, AST function types FunctionType and GenericFunctionType
do not include implicit self in their param list. They represent methods with
implicit self as like: `(Self) -> (Args...) -> Result` and don't have any information
to indicate they may have implicit self. Since we use number of parameters while
deserializing lifetime dependencies, we go wrong for such function types.
Serialize the length of parameter indices, so that lifetime dependencies can be
deserialized to that length.
rdar://151768216
Implements SE-0460 -- the non-underscored version of @specialized.
It allows to specify "internal" (not abi affecting) specializations.
rdar://150033316
By default (currently) the closure passed to a parameter with `@_inheritActorContext`
would only inherit isolation from `nonisolated`, global actor isolated or actor
context when "self" is captured by the closure. `always` changes this behavior to
always inherit actor isolation from context regardless of whether it's captured
or not.
There are four attributes which serialize out a DeclNameRef, sometimes by dropping some of its components. Standardize them with a representation that can handle module selectors.
When serializing `@available` attributes, if the attribute applies to a custom
domain include enough information to deserialize the reference to that domain.
Resolves rdar://138441265.
Followup fix to #80009. We can still get ambiguities from colliding
decls across modules with the deserialization filtering. Bring back
calling the general lookup shadowing after the filtering. This way it
won't use filtered out decls to hide potential candidates.
rdar://148286345
It’s unnecessary, shouldn’t be serialized into module interfaces, and Swift doesn’t know how to compute it for an ABI-only decl since it doesn’t have accessors or an initial value.
No tests because enforcement isn’t in yet.
I am doing this in preparation for adding the ability to represent in the SIL
type system that a function is global actor isolated. Since we have isolated
parameters in SIL, we do not need to represent parameter, nonisolated, or
nonisolated caller in the type system. So this should be sufficient for our
purposes.
I am adding this since I need to ensure that we mangle into thunks that convert
execution(caller) functions to `global actor` functions what the global actor
is. Otherwise, we cannot tell the difference in between such a thunk and a thunk
that converts execution(caller) to execution(concurrent).
The module name changes the symbol mangling, and also causes
TBDGen to emit linker directives. To separate out these two
behaviors, introduce a terrible hack. If the module name
contains a semicolon (`;`), the part before the semicolon
is the module name for mangling, and the part after the
semicolon is the module name for linker directives.
If there is no semicolon, both module names are identical,
and the behavior is the same as before.
An "abstract" ProtocolConformanceRef is a conformance of a type
parameter or archetype to a given protocol. Previously, we would only
store the protocol requirement itself---but not track the actual
conforming type, requiring clients of ProtocolConformanceRef to keep
track of this information separately.
Record the conforming type as part of an abstract ProtocolConformanceRef,
so that clients will be able to recover it later. This is handled by a uniqued
AbstractConformance structure, so that ProtocolConformanceRef itself stays one
pointer.
There remain a small number of places where we create an abstract
ProtocolConformanceRef with a null type. We'll want to chip away at
those and establish some stronger invariants on the abstract conformance
in the future.
* [CS] Decline to handle InlineArray in shrink
Previously we would try the contextual type `(<int>, <element>)`,
which is wrong. Given we want to eliminate shrink, let's just bail.
* [Sema] Sink `ValueMatchVisitor` into `applyUnboundGenericArguments`
Make sure it's called for sugar code paths too. Also let's just always
run it since it should be a pretty cheap check.
* [Sema] Diagnose passing integer to non-integer type parameter
This was previously missed, though would have been diagnosed later
as a requirement failure.
* [Parse] Split up `canParseType`
While here, address the FIXME in `canParseTypeSimpleOrComposition`
and only check to see if we can parse a type-simple, including
`each`, `some`, and `any` for better recovery.
* Introduce type sugar for InlineArray
Parse e.g `[3 x Int]` as type sugar for InlineArray. Gated behind
an experimental feature flag for now.
Recover from a raw type hidden behind an internal or implementation-only
import by dropping the whole enum when the raw type is unavailable. This
scenario should happen only when looking at non-public decl for indexing or
debugging, or if dependencies somehow changed and left behind a stale
swiftmodule file.
rdar://147091863
Rename the macro UNWRAP to SET_OR_RETURN_ERROR for clarity and invert its
parameters to have a more intuitive order of assignment target and then
expression.
When a Swift function shadows a clang function of the same name, the
assumption was that Swift code would refer only to the Swift one.
However, if the Swift function is `@usableFromInline internal` it can be
called only from the local module and inlined automatically in other
clients. Outside of that module, sources see only the clang function, so
their inlinable code calls only the clang function and ignores the Swift
one. This configuration passed type checking but it could crash the
compiler at inlining the call as the compiler couldn't see the clang
function.
Let's update the deserialization logic to support inlined calls to the
shadowed or the shadower. Typical shadowing is already handled by the
custom deserialization cross-reference filtering logic which looks for
the defining module, scope and whether it's a Swift or clang decl. We
can disable the lookup shadowing logic and rely only on the
deserialization filtering.
rdar://146320871
https://github.com/swiftlang/swift/issues/79801
