Make empty interval comparable by reference

[justas-]

As it is now we have to instantiate I.empty() to compare if the interval is empty:

i1 = I.empty()
i1 == I.empty()
True
i1 is I.empty()
False

In certain applications this can have a significant performance impact. Would it be possible to have a singleton for an empty interval, which would allow checking using "is" operator?

[AlexandreDecan]

It's quite unusual in Python to use identity operator (is) to support this kind of use case; e.g. with lists:

>>> m = []
>>> m is []
False

Why not use i1.is_empty() in your case?

[justas-]

Good point! I did not know about is_empty() and got used to using sympy's singleton EmptyInterval.

[AlexandreDecan]

I hope is_empty() will do the job performances-wise ;-)

I would be very interested to get your feedback on how python-intervals' performances are in practice. My use cases usually do not exceed tens of intervals, so I don't know if there are bottlenecks.

[justas-]

My use case involves simulation with a lot of time intervals. For each set of simulation parameters I calculate ~12 time intervals. Then I check if all of them are valid. If they are - then I calculate one intersection of them all and check again if it is valid (not empty). Compared to sympy I see ~30x improvement using python-intervals.

[AlexandreDecan]

Thank you! That's a good news ;-)

[AlexandreDecan]

I know this issue is quite old and a "workaround" has been found, but since 1.8.0, I.empty() always returns the same instance, so comparing some_interval == I.empty() should be nearly as fast than some_interval.is_empty(). Notice that is operator is still not supported (because not all empty intervals are created using I.empty(), so we cannot ensure that their id(...)'s will be the same!)