It's a trade-off. If you store both endpoints you can continue to think of rows as order-invariant tuples. If you store only one endpoint, you have to impose a meaningful order on the rows in order for them to make sense.
Sure, from a theoretical perspective, but in practice there’s got to be some sort of order at some point even when storing timespans.
Sure, from a theoretical perspective, but in practice there’s got to be some sort of order at some point even when storing timespans.