Cool, whether it's pushing forward (Jan 32 → Feb 1; 23:59:60 → 00:00:00 + 1 d) or saturating (Jan 32 → Jan 31; 23:59:60 → 23:59:59) it'd be good to see a consistent, explainable rule-of-thumb (assuming that there is some modicum of alignment among existing implementations on that).