[libcxx-commits] [PATCH] D128927: [libc++] Always build c++experimental.a

Mon Jul 18 07:58:21 PDT 2022

hans added a subscriber: thakis.
hans added a comment.

After discussing with @thakis a bit, I've reverted this in 09cebfb978def7fa2a4460bca89690f8d3608216 <https://reviews.llvm.org/rG09cebfb978def7fa2a4460bca89690f8d3608216> to bring builds back to a green state while we figure out what's going on and how to fix this.

The fact that Greendragon broke (example build <https://green.lab.llvm.org/green/blue/organizations/jenkins/clang-stage1-RA/detail/clang-stage1-RA/30197/pipeline/>) seems like a strong indication that this would break others as well, and removing libc++ from those bots seems like avoiding rather than fixing the problem.

If it's the case that Clang and libc++ should no longer be built together (as indicated on https://github.com/llvm/llvm-zorg/pull/28) I think we need a better failure mode than "fatal error: 'experimental/memory_resource' file not found", some good documentation updates to explain what people should be doing instead, and time for people to fix their builds.

As for Chromium, we've been using the technique suggested by @ldionne in https://bugs.chromium.org/p/chromium/issues/detail?id=1067216#c7 plus `-DLIBCXX_ENABLE_EXPERIMENTAL_LIBRARY=OFF`. We would be grateful for ideas of what we should do instead moving forward.

(Also, this is probably a stupid question, but why can't this library use its own headers? I really don't understand the failure mode here to begin with.)

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D128927/new/

https://reviews.llvm.org/D128927