Currently, this is staged in as `_forget`,
as part of SE-390. It can only be used on
`self` for a move-only type within a consuming
method or accessor. There are other rules, see
Sema for the details.
A `forget self` really just consumes self and
performs memberwise destruction of its data.
Thus, the current expansion of this statement
just reuses what we inject into the end of a
deinit.
Parsing of `forget` is "contextual".
By contextual I mean that we do lookahead to
the next token and see if it's identifier-like.
If so, then we parse it as the `forget` statement.
Otherwise, we parse it as though "forget" is an
identifier as part of some expression.
This way, we won't introduce a source break for
people who wrote code that calls a forget
function.
This should make it seamless to change it from
`_forget` to `forget` in the future.
resolves rdar://105795731
And use the new bit to ensure we don't try to lower move-only types
with common layout value witness surrogates. Take a bit in the runtime
value witness flags to represent types that are not copyable.
rdar://105837040
* WIP: Store layout string in type metadata
* WIP: More cases working
* WIP: Layout strings almost working
* Add layout string pointer to struct metadata
* Fetch bytecode layout strings from metadata in runtime
* More efficient bytecode layout
* Add support for interpreted generics in layout strings
* Layout string instantiation, take and more
* Remove duplicate information from layout strings
* Include size of previous object in next objects offset to reduce number of increments at runtime
* Add support for existentials
* Build type layout strings with StructBuilder to support target sizes and metadata pointers
* Add support for resilient types
* Properly cache layout strings in compiler
* Generic resilient types working
* Non-generic resilient types working
* Instantiate resilient type in layout when possible
* Fix a few issues around alignment and signing
* Disable generics, fix static alignment
* Fix MultiPayloadEnum size when no extra tag is necessary
* Fixes after rebase
* Cleanup
* Fix most tests
* Fix objcImplementattion and non-Darwin builds
* Fix BytecodeLayouts on non-Darwin
* Fix Linux build
* Fix sizes in linux tests
* Sign layout string pointers
* Use nullptr instead of debug value
The `deinit` takes full responsibility for destroying the value, using the
user-defined deinit body and implicitly destroying any remaining resources
not consumed during the deinit.
Remaining to do after this patch:
- Run the deinit for enums
- Pass generic arguments for generic move-only types
- Handle deinits that take their parameter indirectly
- Teach value witness layout about when types are move-only and/or have
deinits, so that we don't attempt to give move-only types standard
value witness tables or share box metadata with copyable payloads
The Swift Simplification pass can do more than the old MandatoryCombine pass: simplification of more instruction types and dead code elimination.
The result is a better -Onone performance while still keeping debug info consistent.
Currently following code patterns are simplified:
* `struct` -> `struct_extract`
* `enum` -> `unchecked_enum_data`
* `partial_apply` -> `apply`
* `br` to a 1:1 related block
* `cond_br` with a constant condition
* `isConcrete` and `is_same_metadata` builtins
More simplifications can be added in the future.
rdar://96708429
rdar://104562580
These were disabled becuase they didn’t work properly in bots testing iOS or other platforms. Re-enable them for macOS until they can be fully repaired.
Type wrapper synthesis cannot reply on `getExecutableInit` to
determine whether initialization should use backing wrapper
call or not, it has to check whether the init expression exists.
Closure cycles were originally enforced "conservatively". Real code
does, however, use recursive local functions. And it's surprisingly
easy to create false exclusivity violations in those cases.
This is safe as long as we can rely on SILGen to keep captured
variables in boxes if the capture escapes in any closure that may call
other closures that capture the same variable.
Fixes https://github.com/apple/swift/issues/61404
Dynamic exclusivity checking bug with nested functions. #61404
Fixes rdar://102056143 (assertion failure due to exclusivity checker -
AccessEnforcementSelection should handle recursive local captures)
If the replaced symbol goes away in the original library, the
replacement key in the replacement descriptor will be null. Handle this
by ignoring the replacement entry rather than crashing.
rdar://103307821
AnyKeyPath's debugDescription assumes there's always at least one component, but `\Type.self` produces an empty keypath. Special-case the empty case to display a `.self` component.
rdar://103237845
An early approach to codegen for `#_hasSymbol` relied on the Darwin platfom SDK, but now that the feature lowers directly to NULL checks in LLVM IR a platform restriction is no longer needed.
However, the tests for `#_hasSymbol` remain unsupported on Windows since that OS does not support weak linking.
- Infer type wrappers only from direct (declared on type) protocols
- Inferences from protocol to protocol is not allowed
- If type specifies a type wrapper attribute explicitly it has to
match the one associated with a declared protocol, otherwise the
declaration is going to be rejected as having multiple wrappers.
Now that the public declarations in API libraries are checked according to the minimum possible deployment target of their clients this relaxation is needed for source compatibility with some existing code.
Resolves rdar://100904631
I added code to the type checker that both bans this when one adds the
conformance on the nominal type declaration as well as on extensions. I was
careful to ensure we can still add extensions without an inherited clause.
Doing this until we get the full generics model.
rdar://101874019
For each decl that needs a `#_hasSymbol()` query function, emit the corresponding helper function body during IRGen. Use `IRSymbolVisitor` to collect linkable symbols associated with the decl and return true from the helper function if the address of every associated symbol is non-null.
Resolves rdar://101884587
Without this change, an `@_objcImplementation` cannot override parent class methods, because the special access control behavior breaks the access control checks for overrides.
