https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65641
--- Comment #7 from Jonathan Wakely <redi at gcc dot gnu.org> --- (In reply to Jakub Lopuszanski from comment #6) > Then the check wouldn't have to use modulo, just &1. > Addresses are aligned at least to multiple of word length, Not on byte-aligned targets such as cris-elf > B) Check if the address is within the buckets array bounds > > As the design dictates that pointers which point to the first items are > stored directly in the array, and others lie outside, it should be possible > to identify that a given pointer is the last one/first, by doing comparison > of the address to the range of addresses occupied by the array. This is not currently possible in constexpr (and C++26 requires the unordered containers to be usable in constexpr).
