When dealing with explicit coercions apply their conversions
early to avoid searching for solutions with incorrect types,
and therefore prune search space.
Move disjunction selection logic one level up from the `solveSimplified`
which allows to simplify its logic and avoid collecting disjunctions
multiple times for each solver step.
Sort constraint components/buckets based on how many disjunctions
they have, that helps to prune some of the branches with incorrect
solutions early which limits overall depth of the search.
Resolves: SR-4714
If we allow the right-hand type to be bound first, it can artificially
limit the options for the left-hand side, so attempt to delay binding of
the right-hand side types until after we've chosen something for the
left-hand side.
The test here will not fail without this change, but will fail without
this change if the ConstraintGraph.cpp changes from @xedin's edge
contraction patch (https://github.com/apple/swift/pull/11118) are
applied.
While shrinking we have to allocate containers for the reduced domains
for some of the candidates, it's currently done using permanent arena
of the `ASTContext` allocator. This patch changes candidate solver to
use arena associated with the parent constraint system, which significantly
limits lifetime of domain containers.
Currently we have a number of unsolved disjunctions hard-coded to 5,
which breaks some existing code by terminating shrinking too early.
This patch makes it a command-line option so users have control over
what that threshold can be.
Resolves: rdar://problem/33433595
The distinction was made between these two because of rdar://25341015,
wherein we needed a diagnostic that could detect the shadowing of
global functions by calls in protocols that don't match any
of the protocol's members.
This used to function by hoping that we had an argument tuple type lying
around after type checking. The expr cleaner would re-write that
type into the AST and we would look up based on it. Now that we write
Type() into the AST on failure, we must instead type check the
component parts of the argument themselves to form a tuple
type to make the lookup.
When we track which expressions we're already in the middle of type
checking, we need to ensure that we also track the constraint system
which has the types for that expression, and then transfer those types
into our current constraint system so that we do not fail to look them up.
Split out the code for selecting potential bindings into a separate file
as a first step before refactoring it for improved clarity and ease of
modification.
In some situations e.g. while trying to shrink domains of the type
variables before attempting search, use a flag to tell constraint
system to retain all of the viable solutions otherwise solver could
loose some of the information required to produce complete solution.
Resolves: rdar://problem/32726044
By default, end expression type checking after the elapsed process time
is more than 60 seconds for the current expression. This threshold can
be overridden by using -solver-expression-time-threshold=<seconds>.
Resolves rdar://problem/32859654
Calls involving single trailing closure arguments require special
handling because we don't have as much contextual information
about function/argument types as in with regular calls, which means
that diagnosing such situations only by `visitApplyExpr`
yields subpar results.
Resolves: SR-4836.
AnyFunctionType::Param carries around information about decomposed
parameters now. Information about default arguments must be computed
separately with swift::computeDefaultMap.
Track outcomes of `conformsToProtocol` calls in `simplifyConformanceConstraints`
to be able to validate conformances when solution is formed to avoid returning
solutions with nominal types with invalid conformances to protocols.
Instead of having three different scores for - string, array, inout,
let's unify them under a single value-to-pointer score and use
it in appropriate places when simplifying restricted constraints.
There are possible situations when we find solutions with String
and String -> UnsafePointer conversions at the same time for
expressions with default string literals. In order to disambiguite
such situations let's prefer solutions without String -> UnsafePointer
conversions if possible.
Restrict skipping of the generic overloads only to the situations
when non-generic solution doesn't have any restrictions/fixes, because
there is a possibility that generic overload could produce a better
solution.
Resolves: rdar://problem/32204609.
Add additional checking for complexity of the shrinking candidate
given the number of the expressions solved so far and total number
of disjunctions present. This allows us to bail quicker in complex
expression cases which, at the very least, produces an error instead
of being "stuck" in solver for a long time.
Resolves: rdar://problem/32034560
More updates to read and write types from a side table in the constraint
solver and only write back into expressions when we've finished type
checking an expression.
We still write directly into expressions, and will continue to do so
until all of the code has been updated.
When this score kind was added, the marker for the last score kind was
not updated. As a result, increasing the score for SK_KeyPathSubscript
was not actually having an effect.
Add tests that include a type with subscripts that also take the key
paths as their argument with an identical argument label ("keyPath").
These tests show that we're actually falling back on the keypath
application in many cases despite the score kind specific to keypath
application. I'll open a couple new JIRAs to investigate this
behavior.
The update to SK_LastScoreKind results in fixing a crash in these tests
that happens as a result of an assumption in CSRanking.cpp that if we're
comparing two solutions we end up with decls for overloads to compare,
which isn't the case here due to the keypath application.
Consolidate some of the attributes and operations common to disjunction choice
in new `DisjunctionChoice` class, which simplifies implementation of the
`ConstraintSystem::solveSimplified` method related to overload selection.
The old TVO_MustBeMaterializable is now equivalent to
!TVO_CanBindToLValue && !TVO_CanBindToInOut.
I tried to update all usages of createTypeVariable() to
pass TVO_CanBindToInOut unless they explicitly passed
TVO_MustBeMaterializable before.
However, in reality TVO_CanBindToInOut is the rare case;
we can remove this flag gradually over time to fix
crashes and diagnostics.
If a lazy var has no declared type, we have to type check the
initializer to get a type before we can build the getter.
Then, the initializer is type checked as part of the getter
again.
Use the new SkipApplyingSolution flag when type checking for
the first time. We still end up doing redundant work, but by
not applying the solution we avoid feeding invalid AST nodes
back into the constraint solver.
This fixes some bad diagnostics and crashes.
Fixes <https://bugs.swift.org/browse/SR-2616> and
<rdar://problem/28313602>.
The openType() function did two things:
- Replace unbound generic types like 'G' with fresh type variables
applied to bound generic types, like 'G<$T0>'.
- Replace generic parameters with type variables from the replacement
map.
The two behaviors were mutually exclusive and never used from the
same call site, so split them up into two independent functions.
Also, eliminate ConstraintSystem::openBindingType() since it was only
used in one place and could be expressed in terms of existing functions.
If we had an unbound generic type whose parent was a bound
generic type, and the outer type's generic signature
placed generic constraints on outer parameters, we would
create type variables that weren't ever bound to anything.
Previously, this would never come up, but it will once
preCheckExpression() is folding more MemberRefExprs down
to TypeExprs.
