Fix gnu_cxx test on GCC. Remove CFI tests which are unnecessary according to EricWF (after offline discussion).

Fix XFAIL for real.

Fix CI issues.

Thanks a lot for picking up this patch! I have a bunch of comments but this is starting to look pretty good.

XFAIL test on Windows

Adjust ignore_format.txt

This looks pretty clean, a lot cleaner than I would have expected for an embedded c library. I have a few questions and comments but I don't see major blockers at this point.

Update after discussion with Eric. This allows fixing the UB in all standard modes and simplifies the code quite a bit!

LGTM with comments applied and CI passing.

@mstorsjo Do you know what's the state of things for compiler-rt on Windows?

This is now https://github.com/llvm/llvm-project/pull/67224

LGTM with minor nitpick. You can ignore the CI failures since they are unrelated and have been fixed since then. Thanks!

Format added test.

Update ignore_format.txt

Heads up for vendors. I'll ship this at the end of the week if nobody chimes in. If this causes problems after being landed, there's a few ways to try and reduce the scope of breakage.

Update patch message.

Rebase the patch, avoid ABI break, make 100% conforming (at the cost of potential breakage).

[Github PR transition cleanup]

Go over all the tests and fix some refactorings that weren't the way they should.

Fix formatting.

Rebase and apply comments. I will merge this if CI is green. Per the update in https://github.com/llvm/llvm-project/issues/61254, I don't think this causes any well-formed code to be rejected, and it fixes a real bug (https://github.com/llvm/llvm-project/issues/61202).

In D145376#4203303, @EricWF wrote:

Is this code ODR used? I suspect it is, and so it's not valid, but I'm not sure and I ran into a bunch of different instances of this pattern while testing this change.

template <class T>
decltype(auto) compute_return_type() {
    // ...
    // stuff that makes this function look reasonable...
    // ...
    return std::declval<T>();
}

using MyT = decltype(compute_return_type<int>());

[Github PR Transition Cleanup]

Fix C++03 build.

Add missing include to test.

Reduce to tuples of size 512, otherwise we don't have enough memory on some builders.

Add missing _LIBCPP_HIDE_FROM_ABI

In D80588#4648327, @EricWF wrote:

A little more digging into this assembly can be found here: https://godbolt.org/z/TrWY7YMWW (or with LLVM IR: https://godbolt.org/z/69xzf7r74)

I think this change is good and safe. Still pondering the "how to test this" question.

@mstorsjo Can you rebase this now? I just dropped support for Clang 15.

Abandoning to clean up the review queue.

Requesting changes since I'd like to discuss the latest changes to the patch during our next 1:1

Rebase and try to fix CI.

[Github PR transition cleanup]

Encode the real version of Clang that we support right now.

Commandeering cause I want to land this.

Blocked on https://github.com/llvm/llvm-project/pull/66824.

Address comments.

Rebase and address comments.

[Github PR transition cleanup]

Rebase. Fix tests and comments.

[Github PR transition cleanup]

[Github PR transition cleanup]

[Github PR transition cleanup]

[Github PR transition cleanup]

Gentle ping. Are we waiting on anything to ship this?

Rebase. Simplify the implementation a bit. Remove some tests for SFINAE-friendliness that are not necessary according to my reading of the spec.

[Github PR transition cleanup]

[Github PR transition cleanup]

In D157283#4614177, @DmitryPolukhin wrote:

@ldionne could you please upstream 9edb9a711503d540cf3126c0fde11ce9a0d9a7aa (I don't know what it is)?
Alternatively I can figure out what to do with this test (it is just copies of the same test with minimal modifications) with a hint how it was resolved to make sure that it worn diverse from want it should be on Darwin.

I just did some digging. We completed the transition to the libcxx-headers-in-the-SDK already, and now upstream and downstream Clang are the same with respect to those search paths. I think it might just be the case that you need the very latest AppleClang for that to be reflected (as you know AppleClang always lags a bit behind). So I think there's nothing special to do here, upstream behaves properly and downstream will catch up automatically.

[Github PR transition cleanup]

Please open as a github PR is you still want to do this (I think it makes sense but we have to drain our review queue here).

[Github PR transition cleanup]