[CodeCompletion] Migrate expression completions to solver-based #41633
Conversation
ahoppen
force-pushed
the
pr/solver-based-global-completions
branch
from
bc8a25a to
f698b2c
Compare
ahoppen
changed the title
ahoppen
changed the title
| SolutionApplicationTarget target(closure, closure, | ||
| ContextualTypePurpose::CTP_CallArgument, | ||
| contextualType, /*isDiscarded=*/false); | ||
| setSolutionApplicationTarget(closure, target); |
There was a problem hiding this comment.
@xedin I’m pretty sure this part is not correct. As we discussed, I’m using SolutionApplicationTarget to convey the contextual type of the ClosureExpr. I’m not entirely sure what you’d expect for the decl context, contextual type purpose and isDiscarded. AFAICT, we don’t really know the contextual type purpose and whether it’s discarded inside resolveClosure.
There was a problem hiding this comment.
I'd suggest you add a new ContextualTypePurpose - CTP_Closure, declaration context of this target should be the one associated with constraint system and isDiscarded is false in here.
| SolutionApplicationTarget target(closure, closure, | ||
| ContextualTypePurpose::CTP_CallArgument, | ||
| contextualType, /*isDiscarded=*/false); | ||
| setSolutionApplicationTarget(closure, target); |
There was a problem hiding this comment.
I'd suggest you add a new ContextualTypePurpose - CTP_Closure, declaration context of this target should be the one associated with constraint system and isDiscarded is false in here.
| /// "resolved" concrete type. | ||
| /// If no type has been computed for this node or if \c node is null, returns | ||
| /// the null type. | ||
| Type getResolvedTypeIfAvailable(ASTNode node) const; |
There was a problem hiding this comment.
I don't see any uses of this but I'd say that if ASTNode doesn't have a type it's either not associated with the solution which is an issue on the caller side, or it didn't get recorded during constraint generation, which makes it a bug in the solver.
There was a problem hiding this comment.
Oh, looks like the need for that method has gone away at some point during my development. I just removed it.
ahoppen
force-pushed
the
pr/solver-based-global-completions
branch
3 times, most recently
from
121603e to
666658e
Compare
ahoppen
force-pushed
the
pr/solver-based-global-completions
branch
from
666658e to
53a677f
Compare
|
@swift-ci Please smoke test |
|
@swift-ci Please SourceKit stress test |
|
Stress tester run: https://ci.swift.org/job/swift-PR-macos-with-sourcekit-stress-tester/44/ This needs some XFails due to SR-16012 and a performance problem I’m addressing in separately in #41917. |
ahoppen
force-pushed
the
pr/solver-based-global-completions
branch
from
53a677f to
e2a62f1
Compare
|
@swift-ci Please smoke test |
- SR-14694: A few timeout issues were resolved by swiftlang/swift#41917 and others were introduced by swiftlang/swift#41633. - SR-16012: A couple more cases were caused by migrating expression completions to solver-based swiftlang/swift#41633
Migrate
StmtOrExprcompletions to solver-based.