Memory unsafety in the iteration part of the for-in loop (i.e., the part
that works on the iterator) can be covered by the "unsafe" effect on
the for..in loop, before the pattern.
The new `DECL_ATTR_FEATURE_REQUIREMENT` macro in DeclAttr.def can be used to declare that an attribute should only be available when a related language feature is enabled.
Effects:
• `#if hasAttribute(someAttr)` will return `false` unless the required feature is enabled.
• Code completion will not include the attribute unless the required feature is enabled.
• `TypeChecker::checkDeclAttributes()` diagnoses non-implicit uses of the attribute.
Add this mechanism and use it to tie @abi to the ABIAttribute feature. Also design tests for it.
The previous logic for this was unused, replace
it with new logic that consults
InvertibleProtocols.def for the list of protocols
to suggest.
rdar://139212286
Add a case for completing type attributes in
inheritance clause position, and limit the
completion of `@unchecked`, `@preconcurrency`, and
`@retroactive` to that case.
Complete ownership specifiers such as `consuming`,
`borrowing`, and `inout` in parameter type
position. While here, also complete `isolated`.
rdar://127261573
When doing solver-based completion, the request
will fail, avoid falling back to a regular call
expression in that case; we only care about
whether we got the callback.
rdar://129024996
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)
Use the generalized implied result logic, and
rename to `isImpliedResult` since that's really
what we're querying here, and it needs to handle
implicit-last-exprs if enabled.
Now that there is an actual declaration for `Copyable` we should be emitting that as a code completion result instead of a keyword, like we currently do.
rdar://109107817
This removes the distinction between argument completions and postfix expr paren completions, which was meaningless since solver-based completion.
It then determines whether to suggest the entire function call pattern (with all argument labels) or only a single argument based on whether there are any existing arguments in the call.
For this to work properly, we need to improve parser recovery a little bit so that it parsers arguments after the code completion token properly.
This should make call pattern heuristics obsolete.
rdar://84809503
The implementation is not 100% perfect but I don’t think it’s worth putting too much effort into it passing more information down in the parser if 'repeat' and 'each' are valid if we are going to remove the current parser anyway.
rdar://95725895
This has the benefit that in case of a code completion crash, we can see what kind of completion crashed without having to symbolicate the stack trace using LLDB.
It also made `trySolverCompletion` small enough that we can inline it into `doneParsing`, which makes debugging easier because you won’t accidentally step over the call to `trySolverCompletion`, which actually performs the completion.
Use FetchContent to include swift-syntax directly in swift. This can be
thought of as an `add_subdirectory` for a directory outside the root.
The default build directory will be `_deps/swiftsyntax-subbuild/`, though
the modules and shared libraries will be built in `lib/swift/host` by
passing down `SWIFT_HOST_LIBRARIES_DEST_DIR` to avoid copying them as we
were doing previously.
These allow multi-statement `if`/`switch` expression
branches that can produce a value at the end by
saying `then <expr>`. This is gated behind
`-enable-experimental-feature ThenStatements`
pending evolution discussion.
There's a subtle stack UAF here - `~CodeCompletionResultBuilder` would
be called *after* `~ExpectedTypeContext` as `ExpectedTypeContext` was
defined after `CodeCompletionResultBuilder`. Fix the order they're
created to prevent this.
This will allow us to run two different completion kinds and deliver results from both of them.
Also: Compute a unified type context for global lookup. Previously, we always used the expected type context of the last lookup. But really, we should be considering all possible types from all constraint system solutions when computing code completion results from the cache.
We didn’t actually have any tests for this. Completions aren’t great here at the moment but since this is an underscored language feature it’s not that important at the moment.
Don't suggest `let` or `var` in e.g the sequence
expression of a `for` loop, or after a `return`.
We ought to do a better job of checking whether
we're in expression position before suggesting
these (as opposed to a pattern), but I'm leaving
that as future work for now.
Reformatting everything now that we have `llvm` namespaces. I've
separated this from the main commit to help manage merge-conflicts and
for making it a bit easier to read the mega-patch.
This is phase-1 of switching from llvm::Optional to std::optional in the
next rebranch. llvm::Optional was removed from upstream LLVM, so we need
to migrate off rather soon. On Darwin, std::optional, and llvm::Optional
have the same layout, so we don't need to be as concerned about ABI
beyond the name mangling. `llvm::Optional` is only returned from one
function in
```
getStandardTypeSubst(StringRef TypeName,
bool allowConcurrencyManglings);
```
It's the return value, so it should not impact the mangling of the
function, and the layout is the same as `std::optional`, so it should be
mostly okay. This function doesn't appear to have users, and the ABI was
already broken 2 years ago for concurrency and no one seemed to notice
so this should be "okay".
I'm doing the migration incrementally so that folks working on main can
cherry-pick back to the release/5.9 branch. Once 5.9 is done and locked
away, then we can go through and finish the replacement. Since `None`
and `Optional` show up in contexts where they are not `llvm::None` and
`llvm::Optional`, I'm preparing the work now by going through and
removing the namespace unwrapping and making the `llvm` namespace
explicit. This should make it fairly mechanical to go through and
replace llvm::Optional with std::optional, and llvm::None with
std::nullopt. It's also a change that can be brought onto the
release/5.9 with minimal impact. This should be an NFC change.
