It appears we need to ignore such illformed data, THANKS FOR LETTING US KNOW GUYS ...

and how the @@@ should we handle these inverted ranges in the long ranges test??

Indeed, and thanks for your message, I've had a nightmare with random tests before seeing it...

There are still inverted ranges, in "Long sets/huge ranges"

can you check now?

[Python] In random tests there are ranges with begin > end. Description says nothing about this ranges.
For example:

[290394, 380460, 510318, 640294, 420228, 490320, 570103, 660426, 240234, 250246, 150312, 150001, 510279, 590378, 590379, 230096]

expected result:

[150312, 150001, 240234, 250246, 290394, 380460, 420228, 490320, 510279, 660426]

(150312 150001) is copied, (590379 230096) is ignored, why?

changed to actually merge the ranges

woops...

added, with performances tests.

done

Duplicate (pretty close but not totally equivalent... -> ?)

In the example test the arguments of assertArrayEquals() should be swapped: expected first, actual second.

Needs random tests

Good catch! Will refactor to account for this edge case, thanks!