[libcxx-commits] [libcxx] [libc++] Make `_Atomic(T)` directly use an alias template (PR #168679)
Louis Dionne via libcxx-commits
libcxx-commits at lists.llvm.org
Thu Jun 18 09:54:11 PDT 2026
ldionne wrote:
I mean, no matter what they *intended*, the wording in the standard is very clear as it stands. And a neutral party reading the Standard and coding against it would conclude that we are non-conforming.
We usually ask for LWG clarification when **what is meant** is unclear. In this case, we're not sure about the reason for that wording, but the wording is entirely unambiguous. I would suggest this:
1. We make this change cause it makes us follow the letter of the standard, which is quite clear.
2. We ask LWG whether they truly intended to force implementations to use an alias. If they didn't and if they meant something else, this turns into a LWG issue that we'll eventually resolve by probably reverting this PR (roughly).
This way, everything's actionable and we unblock landing this PR and fixing the underlying GH issue, and we avoid staying in limbo because we don't think LWG meant what they wrote. WDYT?
https://github.com/llvm/llvm-project/pull/168679
More information about the libcxx-commits
mailing list