The Ventura Core Foundation Merge #4633
Conversation
|
@swift-ci please test |
|
cc @parkera |
|
cc @compnerd - Similar to previous merges, can we kick off a Windows build to see if there's anything breaks? |
|
@iCharlesHu I thought this was the source compiled at Apple for macOS? |
| char asciiBuf[CFMaxPathSize] = {0}; | ||
| int res = WideCharToMultiByte(CP_UTF8, 0, buf, rlen, asciiBuf, sizeof(asciiBuf) / sizeof(asciiBuf[0]), NULL, NULL); | ||
| if (0 < res) { | ||
| char asciiBuf[CFMaxPathSize] = {0}; | ||
| int res = WideCharToMultiByte(CP_UTF8, 0, buf, rlen, asciiBuf, sizeof(asciiBuf) / sizeof(asciiBuf[0]), NULL, NULL); | ||
| if (0 < res) { |
There was a problem hiding this comment.
It is hard to judge because tabs are replaced with spaces, but indentation looks weird here and in some code below.
| @@ -1367,7 +1383,7 @@ CF_PRIVATE Boolean _CFReadMappedFromFile(CFStringRef path, Boolean map, Boolean | |||
| if (0LL == statBuf.st_size) { | |||
| bytes = malloc(8); // don't return constant string -- it's freed! | |||
| length = 0; | |||
| #if TARGET_OS_MAC || TARGET_OS_LINUX || TARGET_OS_BSD || TARGET_OS_WASI | |||
There was a problem hiding this comment.
I see major loss of code under TARGET_OS_WASI and TARGET_OS_BSD. Probably, @MaxDesiatov could point if those are important changes for Swift WASM support. And @3405691582 did most of effort for OpenBSD support.
| @@ -451,290 +447,12 @@ CF_INLINE kern_return_t __CFPortSetRemove(__CFPort port, __CFPortSet portSet) { | |||
| CF_INLINE void __CFPortSetFree(__CFPortSet portSet) { | |||
| close(portSet); | |||
| } | |||
| #elif TARGET_OS_BSD | |||
There was a problem hiding this comment.
I was going to add more finer grained review comments but I just did the needful anyway in #4648 -- feel free to pick those commits on this pr, since I don't know off the top of my head if I can chain prs. The comments about malloc introspection need to be addressed separately though, because it's not immediately apparent how to resolve that issue.
| string = CFStringCreateMutableCopy(allocator, 0, string); | ||
| CFRelease(tmp); | ||
| } | ||
| } else if (string && backingDataForNoCopy && malloc_size((void *)string) != 0) { |
There was a problem hiding this comment.
Avoid unconditional malloc introspection. Can we safely ignore this branch if the platform doesn't support malloc_size?
| string = CFStringCreateMutableCopy(allocator, 0, string); | ||
| CFRelease(tmp); | ||
| } | ||
| } else if (string && backingDataForNoCopy && malloc_size((void *)string) != 0) { |
There was a problem hiding this comment.
Avoid unconditional malloc introspection. Can we safely ignore this branch if the platform doesn't support malloc_size?
This PR brings the FOSS portions of Core Foundation on par with with macOS Ventura, iOS 16, watchOS 9 and tvOS 16. This includes the changes shipped during the previous releases.
Most of the changes include bug fixes and performance improvements, but some are useful to highlight:
CFStringandNSString-based methods, including a new method of generating and storing this information by producing headers rather than bundling binary blobs.CFURLComponentsto support Internationalized Domain Names (IDNs) via Punycode encoding.Some development notes:
CF_CROSS_PLATFORM_EXPORThas been renamed toCF_EXPORT_NONOBJC_ONLYto make its use clearer.os_*_lock_*functions rather thanCFLockorCFMutex. To avoid the codebase diverging too far and make future merges easier, this patch usesos_unfair_lockand similar facilities where they were previously replaced byCFLock. These types shim toCFLockimplementations outside of Darwin.