The missing dependency was revealed by D97572.
Details
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
compiler-rt/test/profile/CMakeLists.txt | ||
---|---|---|
12 | As I've said in https://bugs.chromium.org/p/chromium/issues/detail?id=1194745, I think the right fix would be to make sure that we run these tests against a properly installed copy of Clang + libc++. Instead, what this will do is add a -I path/to/configured/include/c++/v1 argument to the compiler, which should hopefully work but also exposes an implementation detail of libc++'s build (the cxx-headers target) to compiler-rt. TLDR, I think this will work but it's not the ideal fix IMO. |
compiler-rt/test/profile/CMakeLists.txt | ||
---|---|---|
12 | I agree, but we still need to ensure that those headers are in place before we use them so we're still going to need that dependency. |
compiler-rt/test/profile/CMakeLists.txt | ||
---|---|---|
12 | We could do that by ensuring that the compiler-rt build as a whole installs libc++ before running its tests, and then running against the just-installed library (both headers and dylib). But let's do this if it fixes the build now. |
Thanks for the heads up, I accidentally used set(...) instead of list(APPEND ...), it should be addressed by b0d286b03c6e20e57fd2ecf3ea7935669083f3e5.
As I've said in https://bugs.chromium.org/p/chromium/issues/detail?id=1194745, I think the right fix would be to make sure that we run these tests against a properly installed copy of Clang + libc++. Instead, what this will do is add a -I path/to/configured/include/c++/v1 argument to the compiler, which should hopefully work but also exposes an implementation detail of libc++'s build (the cxx-headers target) to compiler-rt.
TLDR, I think this will work but it's not the ideal fix IMO.