I initially assumed this bug was down to a more complex issue around DST changes, but it turned out to be caused just by a naming inconsistency between day and date.
I'm not sure what the history of "date" vs "day" for the unit name is, but certainly in terms of intervals, "day" does make more sense, and I can see why the code that caused this bug looked correct.
I initially assumed this bug was down to a more complex issue around DST changes, but it turned out to be caused just by a naming inconsistency between
day
anddate
.I'm not sure what the history of "date" vs "day" for the unit name is, but certainly in terms of intervals, "day" does make more sense, and I can see why the code that caused this bug looked correct.