Also, stop context type analysis at outer most expression that matches
the position. It seems that that produces more appropriate results.
rdar://problem/30103287
For:
class MyClass {
func foo(x: SomeType)
func foo(x: OtherType)
}
func test(obj: MyClass) {
obj.foo(x: <HERE>)
}
Type checker doesn't keep overloaded choices for 'obj.foo' in the AST
after typechecking. Code completion need to lookup members to collect
possible parameter types.
in unresolved member completion. e.g.
struct Node {
typealias Child = Node
init(children: [Child]) {}
}
let node: Node = .#^COMPLETE^#
This used to emit `.init(children:)` and `.Child(children:)`.
(and TupleExpr at non-call argument position).
Now, unresolved member completion in array literal should work.
Also, Don't calculate convertibility to 'Any' type. That would be a
noise to type relation because anything is convertible to 'Any'.
rdar://problem/43302814
* Handle generic base types
* Suggest '.some' and '.none' for optional types
* Don't look through too many parameter lists for function types
* Include members with convertible type result
rdar://problem/44803439
We need to look through the optional and find the members of T when
doing completion in Optional<T>.
let x: Foo? = .foo
We still don't correctly complete .some/.none, which requires
reconciling the unbound generic type we get from the decl with the real
bound generic type.
rdar://44767478
Using dummy UnresolvedMemberExpr doesn't give us much benefit. Instead, use
CodeCompletionExpr which is type checked as type variable so can use
CodeCompletionTypeContextAnalyzer to infer context types.
This way, we can eliminate most of special logic for UnresolvedMember.
rdar://problem/39098974
Let the solver disregard missing argument labels and record correct
ones, so such problem could be diagnosed later on iff there were no
other more serious failures.
Resetting Lexer to the start of the line result re-tokenizning whole
string literal as a single token, and go past artificial EOF.
This used to end up with assertion failure or infinite-loop in
no-assertion builds.
Quick fix for: rdar://problem/36881302
If we found any error in a list, in most cases, we cannot expect that the
following tokens could construct a valid element. Skip them, instead of trying
to parse them as the next element. This significally reduces bogus diagnostics.
Bailout if seeing tok::eof or token that can never start a element, after
parsing an element. This silences superfluous "expected ',' separator" error,
or misleading expected declaration error. What we should emit is
"expected ')' in expression list, or "expected '}' in struct".
Replace the enum and OptionSet-specific code with static member lookup
that checks the type against the contextual type. This lets us complete
static variables, static functions, and initializers. In particular,
this fixes completion of NS_EXTENSIBLE_STRING_ENUM types from
Objective-C.
rdar://problem/26628652
Fix two crashes related to unresolved-member completion where either the
EnumDecl itself is missing, or its elements have not been type-checked.
Incidentally, resolve the type of the enum elements in the case where I
have observed this happening.
rdar://problem/26860249
The brief explanation is that we are using the type-checker in a
questionable way where for various reasons we can type-check the same
"context" expression more than once. Until we figure out how to stop
doing that, at least avoid this obvious source of issues with ErrorType
showing up during the initial (poor) typecheck and then blocking
progress when we do a more specific check later.
rdar://problem/26462306
rdar://problem/25248190
If a declaration is marked deprecated (but not unavailable) we want to
mark it as "not recommended" so that users know this probably isn't what
you want. We already do something like this in Clang code completions of
deprecated ObjC declarations.
rdar://problem/26335424
The issue here is that the constraint solver was deciding on
FixKind::RelabelCallTuple as the fix for the problem and emitting the
diagnostic, even though there were two different fixes possible.
CSDiags has the infrastructure to support doing doing the right thing
here, but is only being used for ApplyExprs, not SubscriptExprs.
The solution is to fix both problems: remove FixKind::RelabelCallTuple,
to let CSDiags handle the problem, and enhance CSDiags to treat
SubscriptExpr more commonly with ApplyExpr. This improves several cases
where the solver was picking one solution randomly and suggesting that
as a fix, instead of listing that there are multiple different solutions.
Constraint generator likes the input AST to be not type-checked yet. However,
in code completion, we usually feed CSGen a (at least partially)
type-checked AST to infer the types of unresolved members, which
may lead to assertion failures. So we override the asserted method in
the code completion part to avoid explosions.
Swift SVN r32732
Before this commit, for a var decl at top level, code completion engine assumes the code
completion token contained in this decl is the end point of the delayed parsing, i.e. the
second pass of code completion. This may lead to incomplete ASTs during the second pass, for
instance, when the code completion token lies at the second argument of a call expr that takes
three arguments. We fix the issue by allowing the first pass to pre-register the end location of
a top-level var decl since the first pass is a more error-tolerant and holistic parsing, and the second
pass to reuse this loc.
Swift SVN r32729
For completing unresolved members, if constraint solver does not have solutions, we
look up the function names appearing in the same line of the code completion tokens,
and use their parameters that are of enum/optionset type to deliver the code
completion results.
Swift SVN r31256
Before this commit, for unresolved members, code completion suggests all visible enum elements
and option set types. To refine the results, this commit uses constraint solver to infer
the type of unresolved members by analyzing parental expressions. If the solver has solutions,
we complete the unresolved member, otherwise abort.
rdar://16659653
Swift SVN r31195
