Pity that LLVM doesn't get those bitcode improvements used in watchOS and macOS clang is different enough to warrant its own column on cppreference.

Lets see how much Apple Silicon specific optimizations will be upstreamed.

Apple’s strange ABI is upstreamed: https://github.com/llvm/llvm-project/search?q=arm64_32&unsco.... Not sure what “column” you’re talking about, but it’s fairly clear that macOS’s clang is based on a fairly recent LLVM that they branched off of (you can even grab the underlying version number if you know where to look). And profiles for Apple silicon chips are readily available: https://github.com/llvm/llvm-project/blob/954db63cd149df031d.... LLVM is probably one of the projects from Apple that sees the most things get upstreamed.

Mentioned bitcode, not ABI.

Easy to find out which column I mean, https://en.cppreference.com/w/cpp/compiler_support

I did not say that they wouldn't upstream Apple Silicon support, rather that they won't upstream everything, just as most OEMs make with their LLVM forks.