Jacob Bachmeyer wrote: > Perhaps this has already been addressed, but what prevents a 32-bit OS > from nonetheless having a 64-bit time_t? Nothing. That's the case in particular on - Linux on arc, loong32, ork1, riscv32 and x86_64-x32, - NetBSD 6.0 (2012) and later on x86 and sparc, - OpenBSD 5.5 (2014) and later on x86, - FreeBSD/arm, - Minix 3.3 (2014), - MSVC on x86. [1] > Do the tests accommodate the possibility of time_t being long long? Yes, and you find the test results listed in this mail thread [2] under "msvc x86". Bruno [1] gnulib/doc/year2038.texi [2] https://lists.gnu.org/archive/html/autoconf/2023-04/msg00030.html