Should time range checks be inclusive or exclusive?

3
votes

The general convention for a generic range (x,y) is that x is inclusive and y is exclusive.

For Python datetime.time type, datetime.time.max is time(23, 59, 59, 999999), so it doesn't seem to allow to use the conventional range check on the upper end. For example, if I want to check a time range between 10 am and midnight, I might want a range like this: (time(10), time(24)). But time(24) is not valid, even as a sentinel value.

On the other hand, we can't make x exclusive and y inclusive, because then we lose time(0) as a value.

Should range checks on time be inclusive? Something about it doesn't seem right to me, but I can't articulate it.

1
python
If you use datetime instead of time you can use midnight of the following day as your upper bound. But if you don't actually care about the date you would need a dummy value there (1970-1-1 and 1970-1-2, maybe?) which is a bit of a hack. – 0x5453
I notice that the error message for datetime.time(24) is ValueError: hour must be in 0..23. Since 0 and 23 are both permissible values, this may indicate the author's preference about incluvisity of end points in the context of times. – Kevin
I suppose the issue does not really come up in practice, because you would not generally check a time in a finite range of time values. You would check it by using inequalities, so choosing < or <= as you want. For same reason we don't check floats in a range of floats. – wim
Thanks all. I appreciate all the thoughts expressed here. It seems that the practical approach to time ranges depends on the context of the problem. – jacob

1 Answers

0
votes

Use 0 for midnight

Instead of:

(time(10), time(24))

Use:

# From midnight (0:00) to 10am (10:00)
(time(0), time(10))