Add a method to get the size(length) of interval instance.

[hthayko]

e.g.
P.open(1, 2).size() == 1
P.closed(1, 2).size() == 1
(P.open(1, 2) | P.closed(4, 5)).size() == 2

[hthayko]

smth like:

def intervalSize(p):
	ret = 0
	for i in p:
		ret += i.upper - i.lower
	return ret

[AlexandreDecan]

I understand that this would be a nice, simple feature to add to portion. However, it assumes that objects being used as bounds of an interval can be subtracted, which is not an assumption that holds in all possible cases. Therefore, I'm afraid I cannot add this to portion.

Also, I do not agree that P.open(1, 2).size() should return 1 :-) For integers, it should be 0, and for floats, it should be the highest possible float whose value is strictly less than 1.

[hthayko]

Quoting documentation: "Note that discrete intervals are not supported by portion" so I assume the numbered intervals represent sets of real numbers, in which case, the length of an interval is defined as upper - lower, no matter if the ends are open or closed.

Re subtraction being a prerequisite: I agree, it breaks the consistency, but imho the most use cases of portion will be dealing with numbers, and the benefit of having this nice function will make it worth it.

[AlexandreDecan]

Discrete domains are implicitly supported, but not for numbers, since we cannot (always) distinguish between floats and integers (i.e. given [1, 2], does this correspond to integers or real numbers?). For instance, portion works very well with characters, dates, versions (the use case that made me developing this library), etc.

However, if the general definition of "length of an interval" for real numbers is the one you provided, I expect that anyone needing to compute the length of an interval will implement the corresponding function by herself on top of portion. I think it's not a good idea to "break the generalizability" of portion for such a simple function (a potential one-liner would be sum((x.upper - x.lower for x in interval)).

[AlexandreDecan]

I close this issue, but I'll definitely keep the idea in mind for a potential next major release of portion (I'm currently thinking of allowing to create "specialized"/typed intervals in complement to the generic ones, allowing users to provide custom domains/comparisons/operations/infinities).