[5.0] statically map CF types in the ClangImporter #20840
Conversation
As per https://forums.swift.org/t/llp64-targets-and-integral-types/18253/15, define mappings for `CFTypeID`, `CFOptionFlags`, `CFHashCode` to `UInt` for LLP64 support.
|
@swift-ci please test |
|
This should be low risk. This doesn't change any thing in the ABI for Darwin (they should map over as they currently do), but will impact Windows x86_64 support. This ensures that the types are mapped correctly. |
There was a problem hiding this comment.
This is a cherry-pick of #20717. I agree with Saleem's analysis; in fact I'd go even further and say there's no effect on either source or binary compatibility for either Apple platforms or Linux.
|
@swift-ci please test |
|
Build failed |
|
Linux failure is a nondeterministic Foundation XCTest segfault, which I've seen on other PRs. @millenomi, do you have a bug for this already? @swift-ci Please test Linux |
|
@swift5-branch-managers (@tkremenek) ping? |
|
@apple/swift5-branch-managers works |
|
@jrose-apple ... unfortunately, it doesn't work for me (you need to be part of the apple organization to do that since the alias is private) |
|
Oh, I didn't realize! |
As per https://forums.swift.org/t/llp64-targets-and-integral-types/18253/15, define mappings for
CFTypeID,CFOptionFlags,CFHashCodetoUIntfor LLP64 support.Replace this paragraph with a description of your changes and rationale. Provide links to external references/discussions if appropriate.
Resolves SR-NNNN.