Additional restrictions on when protocol inference could be considered
viable. If the chain is connected directly to a contextual type there
cannot be any other inference sources for the base type.
Resolves: rdar://148256978
This PR refactors the ASTDumper to make it more structured, less mistake-prone, and more amenable to future changes. For example:
```cpp
// Before:
void visitUnresolvedDotExpr(UnresolvedDotExpr *E) {
printCommon(E, "unresolved_dot_expr")
<< " field '" << E->getName() << "'";
PrintWithColorRAII(OS, ExprModifierColor)
<< " function_ref=" << getFunctionRefKindStr(E->getFunctionRefKind());
if (E->getBase()) {
OS << '\n';
printRec(E->getBase());
}
PrintWithColorRAII(OS, ParenthesisColor) << ')';
}
// After:
void visitUnresolvedDotExpr(UnresolvedDotExpr *E, StringRef label) {
printCommon(E, "unresolved_dot_expr", label);
printFieldQuoted(E->getName(), "field");
printField(E->getFunctionRefKind(), "function_ref", ExprModifierColor);
if (E->getBase()) {
printRec(E->getBase());
}
printFoot();
}
```
* Values are printed through calls to base class methods, rather than direct access to the underlying `raw_ostream`.
* These methods tend to reduce the chances of bugs like missing/extra spaces or newlines, too much/too little indentation, etc.
* More values are quoted, and unprintable/non-ASCII characters in quoted values are escaped before printing.
* Infrastructure to label child nodes now exists.
* Some weird breaks from the normal "style", like `PatternBindingDecl`'s original and processed initializers, have been brought into line.
* Some types that previously used ad-hoc dumping functions, like conformances and substitution maps, are now structured similarly to the dumper classes.
* I've fixed the odd dumping bug along the way. For example, distributed actors were only marked `actor`, not `distributed actor`.
This PR doesn't change the overall style of AST dumps; they're still pseudo-S-expressions. But the logic that implements this style is now isolated into a relatively small base class, making it feasible to introduce e.g. JSON dumping in the future.
Member chains with leading-dot syntax can infer the base type
only from context, so optionality mismatch with the contextual
type should propagate object type down the chain.
If subtyping is allowed for a result type of an implicit member chain,
let's delay binding it to an optional until its object type resolved too or
it has been determined that there is no possibility to resolve it.
Otherwise we might end up missing solutions since it's allowed to
implicitly unwrap base type but it can't be done early - type variable
representing chain's result type has a different l-valueness comparing
to generic parameter of an optional.
This used to work before due to a hack in constraint generator where
unresolved member with arguments would return base type (which
doesn't allow l-value) instead of a result type.
Resolves: rdar://problem/68094328
We previously were not properly handling the diagnostics for using an rvalue implicit member on the left hand side of an assignment. This adds the proper handling and extends it for member chains.
Instead of creating the type variable for the unresolved member chain at the location of the last member, we now create it at the associated UnresolvedMemberChainResultExpr.
Previously, when the final element of a chain was a ForceValueExpr, the chain result type got caught up in the logic used to allow ForceValueExprs to properly infer lvalue types. By separating the result type variable from the last member of the chain, we make sure to keep that logic focused only on ForceValueExpr.
when we have an optional type. This uncovered an error with unresolved member lookup where we allowed an unresolved value member constraint to fail if lookup failed in an optional type wrapping a type variable.
This resolves SR-13357.
Previously we could skip default literal or
supertype bindings if we had already found a solution
with fixes, which could lead us to miss bindings
that produce better diagnostics.
Tweak the logic such that we continue exploring if
we're in diagnostic mode.
Resolves SR-12399.
automatically.
This commit also renames `ConstraintSystem::recordHole/isHole` to
`recordPotentialHole` and `isPotentialHole` to make it clear that
we don't know for sure whether a type variable is a hole until it's
bound to unresolved.
Introduce a fix to detect and diagnose situations when omitted
generic arguments couldn't be deduced by the solver based on
the enclosing context.
Example:
```swift
struct S<T> {
}
_ = S() // There is not enough context to deduce `T`
```
Resolves: rdar://problem/51203824
* Make sure that base and unwrapped types aren't null
* Don't allocate `ForceOptional` fix if nothing has been unwrapped
in `simplifyApplicableFnConstraint`
* Add sugar reconstitution support to `FailureDiagnostic::resolveType`
* Format diagnostics a bit better
Parse 'var [behavior] x: T', and when we see it, try to instantiate the property's
implementation in terms of the given behavior. To start out, behaviors are modeled
as protocols. If the protocol follows this pattern:
```
protocol behavior {
associatedtype Value
}
extension behavior {
var value: Value { ... }
}
```
then the property is instantiated by forming a conformance to `behavior` where
`Self` is bound to the enclosing type and `Value` is bound to the property's
declared type, and invoking the accessors of the `value` implementation:
```
struct Foo {
var [behavior] foo: Int
}
/* behaves like */
extension Foo: private behavior {
@implements(behavior.Value)
private typealias `[behavior].Value` = Int
var foo: Int {
get { return value }
set { value = newValue }
}
}
```
If the protocol requires a `storage` member, and provides an `initStorage` method
to provide an initial value to the storage:
```
protocol storageBehavior {
associatedtype Value
var storage: Something<Value> { ... }
}
extension storageBehavior {
var value: Value { ... }
static func initStorage() -> Something<Value> { ... }
}
```
then a stored property of the appropriate type is instantiated to witness the
requirement, using `initStorage` to initialize:
```
struct Foo {
var [storageBehavior] foo: Int
}
/* behaves like */
extension Foo: private storageBehavior {
@implements(storageBehavior.Value)
private typealias `[storageBehavior].Value` = Int
@implements(storageBehavior.storage)
private var `[storageBehavior].storage`: Something<Int> = initStorage()
var foo: Int {
get { return value }
set { value = newValue }
}
}
```
In either case, the `value` and `storage` properties should support any combination
of get-only/settable and mutating/nonmutating modifiers. The instantiated property
follows the settability and mutating-ness of the `value` implementation. The
protocol can also impose requirements on the `Self` and `Value` types.
Bells and whistles such as initializer expressions, accessors,
out-of-line initialization, etc. are not implemented. Additionally, behaviors
that instantiate storage are currently only supported on instance properties.
This also hasn't been tested past sema yet; SIL and IRGen will likely expose
additional issues.
- Improve the specific cases of nil and empty collection literals.
- Improve cases of contextual member lookup where the result type of the looked up member disagrees with context.
- Add some fixme's to the testsuite for cases of this diagnostic that should be diagnosed in other ways.
Type resolution wasn't looking through property initializer decl contexts
to find out whether an unbound generic type reference was referring to the
enclosing type. Previously we'd reject this with:
error: cannot convert value of type 'Int' to specified type 'Int'
private var data: Int = Matrix4.size()
~~~~~~~~^~~~~~
which was super confusing. The problem was that we weren't resolving
Matrix4 to Matrix4<T>.
a constraint system in "allowFreeTypeVariables" mode. Previously, we
only allowed a few specific constraints, now we allow any relational and
member constraints. The later one is a big deal because it means that we
can allow ".Foo" expressions as ambiguous solutions, which CSDiags can
handle well.
This unblocks solving 23942743 and enables some minor improvements across
the board, including diagnosing things like this better:
Optional(.none) // now: generic parameter 'T' could not be inferred
That said, it also just permutes some non-awesome diagnostics.
where we type check the destination first, then apply its type to the source.
This allows us to get diagnostics for assignments that are as good as PBD
initializers and other cases.
Swift SVN r31404
diagnoseGeneralFailure to be named diagnoseConstraintFailure and change how
it works:
Now it ranks unresolved constraints in the system based on kind (e.g. whether
they are favored, member constraints ahead of conversion constraints, etc) and
then tries to emit a diagnostic for each failure kind one after another.
This means that if there are multiple failed conversion constraints, but one
is obviously satisfiable, that we continue on to diagnose the next one. This
clears up a swath of embarassing diagnostics and refixes:
<rdar://problem/19658691> QoI: Incorrect diagnostic for calling nonexistent members on literals
Swift SVN r31046
"unavoidable failure" path, along with Failure::DoesNotHaveNonMutatingMember and
just doing some basic disambiguation in CSDiags.
This provides some benefits:
- Allows us to plug in much more specific diagnostics for the existing "only has
mutating members" diagnostic, including producing notes for why the base expr
isn't mutable (see e.g. test/Sema/immutability.swift diffs).
- Corrects issues where we'd drop full decl name info for selector references.
- Wordsmiths diagnostics to not complain about "values of type Foo.Type" instead
complaining about "type Foo"
- Where before we would diagnose all failures with "has no member named", we now
distinguish between when there is no member, and when you can't use it. When you
can't use it, you get a vauge "cannot use it" diagnostic, but...
- This provides an infrastructure for diagnosing other kinds of problems (e.g.
trying to use a private member or a static member from an instance).
- Improves a number of cases where failed type member constraints would produce uglier
diagnostics than a different constraint failure would.
- Resolves a number of rdars, e.g. (and probably others):
<rdar://problem/20294245> QoI: Error message mentions value rather than key for subscript
Swift SVN r30715
get the same wording, fixing <rdar://problem/21964599> Different diagnostics for the same issue
While I'm in the area, remove some dead code.
Swift SVN r30713
When a line begins with '.', it's almost always due to a method chain, not an attempt to start an expression with a contextual member lookup. This is a more principled grammar rule than the long tail of hacks we've been putting up to try to accommodate "builder pattern" usages. Fixes rdar://problem/20238557.
Swift SVN r29606
Most tests were using %swift or similar substitutions, which did not
include the target triple and SDK. The driver was defaulting to the
host OS. Thus, we could not run the tests when the standard library was
not built for OS X.
Swift SVN r24504
rdar://problem/17198298
- Allow 'static' in protocol property and func requirements, but not 'class'.
- Allow 'static' methods in classes - they are 'class final'.
- Only allow 'class' methods in classes (or extensions of classes)
- Remove now unneeded diagnostics related to finding 'static' in previously banned places.
- Update relevant diagnostics to make the new rules clear.
Swift SVN r24260
These changes make the following improvements to how we generate diagnostics for expression typecheck failure:
- Customizing a diagnostic for a specific expression kind is as easy as adding a new method to the FailureDiagnosis class,
and does not require intimate knowledge of the constraint solver’s inner workings.
- As part of this patch, I’ve introduced specialized diagnostics for call, binop, unop, subscript, assignment and inout
expressions, but we can go pretty far with this.
- This also opens up the possibility to customize diagnostics not just for the expression kind, but for the specific types
involved as well.
- For the purpose of presenting accurate type info, partially-specialized subexpressions are individually re-typechecked
free of any contextual types. This allows us to:
- Properly surface subexpression errors.
- Almost completely avoid any type variables in our diagnostics. In cases where they could not be eliminated, we now
substitute in "_".
- More accurately indicate the sources of errors.
- We do a much better job of diagnosing disjunction failures. (So no more nonsensical ‘UInt8’ error messages.)
- We now present reasonable error messages for overload resolution failures, informing the user of partially-matching
parameter lists when possible.
At the very least, these changes address the following bugs:
<rdar://problem/15863738> More information needed in type-checking error messages
<rdar://problem/16306600> QoI: passing a 'let' value as an inout results in an unfriendly diagnostic
<rdar://problem/16449805> Wrong error for struct-to-protocol downcast
<rdar://problem/16699932> improve type checker diagnostic when passing Double to function taking a Float
<rdar://problem/16707914> fatal error: Can't unwrap Optional.None…Optional.swift, line 75 running Master-Detail Swift app built from template
<rdar://problem/16785829> Inout parameter fixit
<rdar://problem/16900438> We shouldn't leak the internal type placeholder
<rdar://problem/16909379> confusing type check diagnostics
<rdar://problem/16951521> Extra arguments to functions result in an unhelpful error
<rdar://problem/16971025> Two Terrible Diagnostics
<rdar://problem/17007804> $T2 in compiler error string
<rdar://problem/17027483> Terrible diagnostic
<rdar://problem/17083239> Mysterious error using find() with Foundation types
<rdar://problem/17149771> Diagnostic for closure with no inferred return value leaks type variables
<rdar://problem/17212371> Swift poorly-worded error message when overload resolution fails on return type
<rdar://problem/17236976> QoI: Swift error for incorrectly typed parameter is confusing/misleading
<rdar://problem/17304200> Wrong error for non-self-conforming protocols
<rdar://problem/17321369> better error message for inout protocols
<rdar://problem/17539380> Swift error seems wrong
<rdar://problem/17559593> Bogus locationless "treating a forced downcast to 'NSData' as optional will never produce 'nil'" warning
<rdar://problem/17567973> 32-bit error message is really far from the mark: error: missing argument for parameter 'withFont' in call
<rdar://problem/17671058> Wrong error message: "Missing argument for parameter 'completion' in call"
<rdar://problem/17704609> Float is not convertible to UInt8
<rdar://problem/17705424> Poor error reporting for passing Doubles to NSColor: extra argument 'red' in call
<rdar://problem/17743603> Swift compiler gives misleading error message in "NSLayoutConstraint.constraintsWithVisualFormat("x", options: 123, metrics: nil, views: views)"
<rdar://problem/17784167> application of operator to generic type results in odd diagnostic
<rdar://problem/17801696> Awful diagnostic trying to construct an Int when .Int is around
<rdar://problem/17863882> cannot convert the expression's type '()' to type 'Seq'
<rdar://problem/17865869> "has different argument names" diagnostic when parameter defaulted-ness differs
<rdar://problem/17937593> Unclear error message for empty array literal without type context
<rdar://problem/17943023> QoI: compiler displays wrong error when a float is provided to a Int16 parameter in init method
<rdar://problem/17951148> Improve error messages for expressions inside if statements by pre-evaluating outside the 'if'
<rdar://problem/18057815> Unhelpful Swift error message
<rdar://problem/18077468> Incorrect argument label for insertSubview(...)
<rdar://problem/18079213> 'T1' is not identical to 'T2' lacks directionality
<rdar://problem/18086470> Confusing Swift error message: error: 'T' is not convertible to 'MirrorDisposition'
<rdar://problem/18098995> QoI: Unhelpful compiler error when leaving off an & on an inout parameter
<rdar://problem/18104379> Terrible error message
<rdar://problem/18121897> unexpected low-level error on assignment to immutable value through array writeback
<rdar://problem/18123596> unexpected error on self. capture inside class method
<rdar://problem/18152074> QoI: Improve diagnostic for type mismatch in dictionary subscripting
<rdar://problem/18242160> There could be a better error message when using [] instead of [:]
<rdar://problem/18242812> 6A1021a : Type variable leaked
<rdar://problem/18331819> Unclear error message when trying to set an element of an array constant (Swift)
<rdar://problem/18414834> Bad diagnostics example
<rdar://problem/18422468> Calculation of constant value yields unexplainable error
<rdar://problem/18427217> Misleading error message makes debugging difficult
<rdar://problem/18439742> Misleading error: "cannot invoke" mentions completely unrelated types as arguments
<rdar://problem/18535804> Wrong compiler error from swift compiler
<rdar://problem/18567914> Xcode 6.1. GM, Swift, assignment from Int64 to NSNumber. Warning shown as problem with UInt8
<rdar://problem/18784027> Negating Int? Yields Float
<rdar://problem/17691565> attempt to modify a 'let' variable with ++ results in typecheck error about @lvalue Float
<rdar://problem/17164001> "++" on let value could give a better error message
Swift SVN r23782
t2.swift:3:1: error: argument for generic parameter 'U' could not be
inferred
f(i)
^
t2.swift:2:6: note: in call to function 'f'
func f<T, U>(t: T) -> U? { return nil }
^
Our lack of decent locator information means that we don't get notes
in all of the cases we want them. I'll look at that separately.
Swift SVN r21921
Start capitalizing on some of the new diagnostic machinery in a few different ways:
- When mining constraints for type information, utilize constraints "favored" by the overload resolution process.
- When printing type variables, if the variable was created by opening a literal expression, utilize the literal
default type or conformance if possible.
- Utilize syntactic information when crafting diagnostics:
- If the constraint miner can produce a better diagnostic than the recorded failure, diagnose via constraints.
- Factor in the expression kind when choosing which types to include in a diagnostic message.
- Start customizing diagnostics based on the amount of type data available.
What does all this mean?
- Fewer type variables leaking into diagnostic messages.
- Far better diagnostics for overload resolution failures. Specifically, we now print proper argument type data
for failed function calls.
- No more "'Foo' is not convertible to 'Foo'" error messages
- A greater emphasis on type data means less dependence on the ordering of failed constraints. This means fewer
inscrutable diagnostics complaining about 'UInt8' when all the constituent expressions are of type Float.
So we still have a ways to go, but these changes should greatly improve the number of head-scratchers served up
by the type checker.
These changes address the following radars:
rdar://problem/17618403
rdar://problem/17559042
rdar://problem/17007456
rdar://problem/17559042
rdar://problem/17590992
rdar://problem/17646988
rdar://problem/16979859
rdar://problem/16922560
rdar://problem/17144902
rdar://problem/16616948
rdar://problem/16756363
rdar://problem/16338509
Swift SVN r20927
When we see a '.member' expression in optional context, look for the member in the optional's object type if it isn't found in Optional itself. <rdar://problem/16125392>
Swift SVN r19469
