Interesting...
I have only print(v) in my code (Python) and after running full tests I've got:
Passed: 161 Failed: 51 Exit Code: 1
:D :D :D
So 152 test cases are for "Non-existant values" and only 48 for valid "start point".
Many reruns of testing those tests returns similar proportion for passed and failed tests.
Is it ok? I don't think so.
I can imagine, in rare case, that all random test will have "Non-existant values" and then you need to hardcode 9 cases from
"Fixed Test" and voile, you're done.
I'm marking this comment as sugestion to fix those random test.

You can. If you get execption from math maybe is becouse of first test that have 0 as argument and you cannot log 0.

Thanks, of course now my code works fine :)

I have the same.
I think Python 3 creates flot (e.g. 1e17) from longer itegeres and even when we recreate it to int, we loose some accurateness at the end of number.

Brain-melting kata, but now when I've finished it I think I'm loving it :D

Testing include numbers larger than 10000, for exapmle: "Testing for 915953073004621 (1 Passed, 0 Failed)" or "Testing for 167702767613034 (0 Passed, 1 Failed)".
And as You see, i passed for 915953073004621, which is greater than second one, which i faild ->
"It should work for random inputs too: 1.4062109132535725e+28 should equal 1.4062109132535727e+28"

Thery is small difference at the end of number :/

I'm so angry!
I try everything to make it right for python! But with those huge numbers there is always problem with accuracy, like this "1.2485283107040636e+21 should equal 1.2485283107040634e+21".
Any advice, how to solve this problem?