Restore wmemchr optimization in ranges::find by AlexGuteniev · Pull Request #5628 · microsoft/STL (original) (raw)

ranges::find is not using wmemchr after #5284. Yet, std::find uses it.

For wmemchr we drag in <cwchar> which is heavy, specifically due to wmemchr optimization. So it seems silly not to take the full advantage of wmemchr. The PR addresses that by making ranges::find also using wmemchr.

This affects ARM64 (non-EC), and builds with vector algorithms disabled.

See also #5596. Not sure if this PR closes that issue, because merging the PR does not preclude doing also Windows SDK throughput optimization or otherwise addressing the throughput in the future.