Some constraint transformations require knowledge about what state
constraint system is currently in e.g. `constraint generation`,
`solving` or `diagnostics` to make a decision whether simplication
is possible. Notable example is `keypath dynamic member lookup`
which requires a presence of `applicable fn` constraint to retrieve
some contextual information.
Currently presence or absence of solver state is used to determine
whether constraint system is in `constraint generation` or `solving`
phase, but it's incorrect in case of `diagnoseFailureForExpr` which
tries to simplify leftover "active" constraints before it can attempt
type-check based diagnostics.
To make this more robust let's introduce (maybe temporarily until
type-check based diagnostics are completely obsoleted) a proper
notion of "phase" to constraint system so it is always clear what
transitions are allowed and what state constraint system is
currently in.
Resolves: rdar://problem/57201781
The lookupDirect means that we can see type declarations nested inside of a protocol. If we do not filter these invalid declarations, we will offer a bogus fixit on top of a cycle diagnostic. Remove these types from consideration entirely so we don't crash and don't offer bogus fixits.
Resolves rdar://57003317
This commit changes how we represent caller-side
default arguments within the AST. Instead of
directly inserting them into the call-site, use
a DefaultArgumentExpr to refer to them indirectly.
The main goal of this change is to make it such
that the expression type-checker no longer cares
about the difference between caller-side and
callee-side default arguments. In particular, it
no longer cares about whether a caller-side
default argument is well-formed when type-checking
an apply. This is important because any
conversions introduced by the default argument
shouldn't affect the score of the resulting
solution.
Instead, caller-side defaults are now lazily
type-checked when we want to emit them in SILGen.
This is done through introducing a request, and
adjusting the logic in SILGen to be more lenient
with ErrorExprs. Caller-side defaults in primary
files are still also currently checked as a part
of the declaration by `checkDefaultArguments`.
Resolves SR-11085.
Resolves rdar://problem/56144412.
If a protocol requirement has a type that's a nested member
type of another member type, eg,
protocol P {
associatedtype A : Q
func f(_: A.B)
}
Then we don't actually want to use 'f()' to infer the witness
for 'A'. By avoiding doing so, we eliminate some cycles which
can allow some programs to type check that didn't before.
When SE-110 was being implemented, we accidentally began to accept
closure parameter declarations that had no associated parameter names,
e.g.
foo { ([Int]) in /**/ }
This syntax has never been sanctioned by any version of Swift and should
be banned. However, the change was made long enough ago and there are
enough clients relying on this, that we cannot accept the source break
at the moment. For now, add a bit to ParamDecl that marks a parameter
as destructured, and back out setting the invalid bit on the type repr
for these kinds of declarations.
To prevent further spread of this syntax, stub in a warning that offers
to insert an anonymous parameter.
Resolves part of rdar://56673657 and improves QoI for errors like
rdar://56911630
Also, add `-solver-expression-time-threshold=1` to a few tests.
This forces sr139, rdar25866240, and rdar23327871 to be moved from the
"fast" directory to the "slow" directory.
If a property has multiple property wrappers attached, we'll have multiple nested calls, where each call's argument is a call to construct the next wrapper in the chain. However, when we use multiple wrappers consecutively, we cannot just rely on the call's type matching the innermost wrapper's type, because it will match the first wrapper in the sequence of consective wrappers and we'll end up crashing in SILGen. So, we should check if the call's argument is another call and look into that before checking the types.
* [TypeChecker] Enclosing stubs protocol note within editor mode
* [test] Removing note from test where there is no -diagnostics-editor-mode flag
* Formatting modified code
* [tests] Fixing tests under validation-tests
all cases of missing generic parameters.
In `ComponentStep::take` when there are no bindings or disjunctions, use hole
propagation to default remaining free type variables that aren't for generic
parameters and continue solving. Rather than using a defaultable constraint for
holes, assign a fixed type directly when we have no bindings to try.
The non-gyb type checker performance tests tend to require non-assert
and release builds. This makes two tests consistent with the others.
This also fixes sr139.swift, which was needlessly running the type checker twice.
This is because we already emit a diagostic to tell the user that the property's type does not match the wrappedValue type, so this diagnostic can be a bit confusing especially because the initializer is synthesized
Push through an easy refactoring to the way we validate and install
implicit constructors. This patch would be NFC but for a regression
test that now must diagnose. #26159 changed validation order in such
a way that the code in validation-test-macosx-x86_64/compiler_crashers_2_fixed/0124-sr5825.swift
used to be accepted. This patch once again changes validation order, so
we now reject this code, restoring the behavior seen on all prior
versions of Swift.
On its face, this test should work. In order for it to do so, witness
matching has to be smarter about the declarations it asks for their
interface type, or it will risk these circular constructions
accidentally being accepted or rejected on a whim.
