__GCC_CONSTRUCTIVE_SIZE and __GCC_DESTRUCTIVE_SIZE are available since GCC 12. I'm assuming clang will also implement these for compatability with libstdc++.
Details
- Reviewers
ldionne Mordante var-const - Group Reviewers
Restricted Project - Commits
- rG56a33ba3dd2b: [libc++] Implement P0154R1 (Hardware inference size)
Diff Detail
- Repository
- rG LLVM Github Monorepo
Unit Tests
Event Timeline
Do you know whether Clang intends to implement this and if so whether they will use the same approach?
Since GCC 12 hasn't been released yet there's no CI testing done. So I would prefer to wait with this patch until we have GCC 12 in the CI.
libcxx/docs/Status/Cxx17Papers.csv | ||
---|---|---|
38 | I don't think we can claim we're implemented this. ATM it's not implemented in Clang. |
I just pinged this topic: https://discourse.llvm.org/t/rfc-c-17-hardware-constructive-destructive-interference-size/48674/37?u=ldionne
Let's see if someone from Clang jumps on it. In all cases, I really don't like implementing this on GCC only, but if that's a way to exert slight pressure on Clang to implement it, I think it's acceptable.
Requesting changes for now, but I don't see anything wrong with the patch as-is.
Let's rebase this, apply my comments and land this if CI is green.
libcxx/docs/Status/Cxx17Papers.csv | ||
---|---|---|
30 | Let's also add a release note. | |
libcxx/test/std/language.support/support.dynamic/hardware_inference_size.compile.pass.cpp | ||
10 | Can we use // XFAIL: clang && stdlib=libc++ instead of #if defined(__cpp_lib_hardware_interference_size) ...? |
libcxx/docs/Status/Cxx17Papers.csv | ||
---|---|---|
38 | Ugh yeah, that's true, we need to add a note. |
Let's also add a release note.