https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122466
--- Comment #3 from James K. Lowden <jklowden at gcc dot gnu.org> --- If you're referring to st_mtim, yes, agreed. I'll avoid it. Under NOTES in the stat(2) man page for current glibc on Ubuntu, it says "Older kernels ... did not support nanosecond timestamp fields. Instead, there were three timestamp fields—st_atime, st_mtime, and st_ctime—typed as time_t that recorded timestamps with one-second precision." It's my understanding those fields are still supported unconditionally. My intention is avoid the nanosecond complexity an expose only the old-school whole-second members in the COBOL POSIX binding for stat(2). We'll have st_mtime and not st_mtim or st_mtimespec. If in the future someone needs nanoseconds, we can add them then. They're very small.
