On 01/23/2018 05:29 AM, Michael H Deckers wrote:

    The assertion        # ... but there's no        # convenient single notation for the date and time of this transition        # so we must duplicate the Rule lines.     is clearly wrong when zic allows negative times of day with     the u suffix -- but it is still in australasia in 2018b.

Good point. I reworded that in the attached patch, which also cleans up some other stale comments that I discovered while looking into this one. I installed this in the master branch, after 2018c was tagged.

Similarly, the (future) change in meaning for tm_isdst (from an     indicator of daylight saving time to one of non-standard time)     should be applied throughout the whole documentation of tzdb.     For instance, newctime.3.txt says         Tm_isdst is non-zero if summer time is in effect.     and this will probably become false (for the effect of     localtime()) after the change.

It's probably better to consistently use the phrase "daylight saving time", as this is common usage (at least in the US) and it maps well to the tm_isdst variable of C. Logically DST can be observed in summer just as easily as in winter, and there's no logical requirement that it must advance the clock: after all, one can "save" daylight at the start of the day, and this is just as easy as "saving" it at the end.