[CursorInfo] Prefer AST based results over solver based

[bnbarham]

Solver based results are fast within a function, where the ASTContext can be re-used. But it is significantly slower than the AST based results when outside of a function. Instead of using solver based as the primary results, only use them as a fallback for when AST based fails.

Resolves rdar://108930110.

[bnbarham]

@swift-ci please test

[bnbarham]

@swift-ci please Sourcekit stress test

[bnbarham]

preset=asan
@swift-ci please test with preset macOS platform

[bnbarham]

@swift-ci please test Linux platform

[ahoppen]

Thank you

[drodriguez]

FYI the changes in test/SourceKit/CursorInfo/static_vs_class_spelling.swift fail when compiling without assertions. I have not been able to figure out exactly where, but when trying to get the static or the class spelling in SwiftLangSupport.cpp UIdentVisitor::visitVarDecl, when calling getCorrectStaticSpelling, in the case of the compiler with assertions getParentPatternBinding returns a value, while in the without assertions case, it returns nullptr. I have been trying to find where setParentPatternBinding might not be called depending on assertions/no assertions and I cannot find it.

[bnbarham]

FYI the changes in test/SourceKit/CursorInfo/static_vs_class_spelling.swift fail when compiling without assertions.

That's... very strange. Is that a clean build? I don't have a release build handy, but checking the response from an actual release sourcekitd does give static as well (that's not checking main though, so it's possible something has changed).

[drodriguez]

That's... very strange. Is that a clean build? I don't have a release build handy, but checking the response from an actual release sourcekitd does give static as well (that's not checking main though, so it's possible something has changed).

Clean builds on both the asserts and the no-asserts when I started. I needed to switch to use inproc for sourcekitd to be able to debug.

By "actual release sourcekitd" do you mean the one that shipped with some Xcode?

[bnbarham]

By "actual release sourcekitd" do you mean the one that shipped with some Xcode?

Yeah, I just tried on the latest release. It's possible there's something new that's causing this, it was just an extra data point.

[bnbarham]

Thanks @drodriguez. For now I've put up #65809 so that the test is passing. The underlying issue here is that we're not actually deserializing the PBD.