[ConstraintSystem] Refactor disjunction partitioning #19962
Conversation
Change the order of statements but nothing functional.
…types. Move this into a separate function.
Unnest code by splitting it out into lambdas.
|
@swift-ci Please smoke test |
lib/Sema/CSSolver.cpp
Outdated
| designatedNominalTypes[designatedTypeIndex]; | ||
|
|
||
| if (parentDecl == designatedNominal) { | ||
| auto &constraints = definedInDesignatedType[designatedTypeIndex]; |
There was a problem hiding this comment.
I think bodies of both of these ifs could be unified into:
// preamble
auto *parentDC = functionDecl->getParent();
auto *parentDecl = parentDC->getSelfNominalTypeDecl();
// in the body
if (parentDecl == designatedNominal) {
auto &container = parentDC->isExtensionContext() ? definedInExtensionOfDesignatedType : definedInDesignatedType;
container[designatedTypeIndex].push_back(constraint);
return true;
}There was a problem hiding this comment.
Yes, something like this should work.
lib/Sema/CSSolver.cpp
Outdated
| } | ||
| } | ||
| SmallVector<SmallVector<unsigned, 4>, 4> definedInDesignatedType; | ||
| SmallVector<SmallVector<unsigned, 4>, 4> definedInExtensionOfDesignatedType; |
There was a problem hiding this comment.
I'm not sure why would we'd want to pass these into a partitionForDesignatedTypes call instead of just making both local in there because they are not used partitionDisjunction?...
There was a problem hiding this comment.
Oversight. Will fix momentarily.
|
@swift-ci Please smoke test |
|
@rudkx Is this really the reason that Driver/sdk-apple.swift test case is failing now in CI? There aren't any other commits: |
|
@atrick That would be pretty unusual. It looks like another run is happening now so we can see how far that gets. I did a complete |
No description provided.