[cxx-interop] Cast "data" to the correct type in _swift_dispatch_data_apply. #34266
Conversation
The implicit conversions are OK in C but C++ will error. To make this header valid in both C and C++ we should just always cast.
| __swift_shims_dispatch_data_applier SWIFT_DISPATCH_NOESCAPE applier) { | ||
| return dispatch_data_apply(data, ^bool(dispatch_data_t data, size_t off, const void *loc, size_t size){ | ||
| return applier(data, off, loc, size); | ||
| return dispatch_data_apply((dispatch_data_t)data, ^bool(dispatch_data_t data, size_t off, const void *loc, size_t size){ |
There was a problem hiding this comment.
An alternative solution would be to update dispatch_data_apply to take a __swift_shims_dispatch_data_t. That would modify a public function to accept a private type so, I figured this solution would be better.
zoecarver
changed the title
|
@swift-ci test |
|
Build failed |
|
Java error. Unrelated. @swift-ci please test OS X Platform. |
|
@swift-ci please test OS X. |
|
Thanks, @varungandhi-apple. I'll commit this tomorrow unless anyone else wants to review. |
|
It's possible that this [1] build failure was caused by this patch. A (full) OS X test did succeed for this patch before it landed, though. Should I revert or wait until the next build also fails (or doesn't fail). I'll plan on the latter unless I hear otherwise. [1] = https://ci.swift.org/job/oss-swift-incremental-RA-osx/12891/ |
|
I think the failure: is unrelated to this patch. I think we've seen it sometimes non-deterministically, I suspect it is an incremental build issue. |
I think you're right. That makes the most sense. I just saw all the other commits were all lldb stuff and this was a failure while building the stdlib. Thanks for looking into it :) |
3 participants
The implicit conversions are OK in C but C++ will error. To make this header valid in both C and C++ we should just always cast.
This and swiftlang/swift-corelibs-foundation#2895 fix the Linux builds for #32721.