Haskell translation

I've made an attempt at generating random valid inputs. Loops aren't generated due to the complexity of ensuring that they don't loop forever. However, they should serve the purpose of weeding out any hard-coded solutions.

python new test framework + random tests are required. updated in this fork

Could be excellent, but.. returning a const char* is bad c++ practice :-(

tests are not very torough in JavaScript either, I had to add a fixed test to invalidate my solution. It needs more test cases like [857, 85]

c++ version should use std::string instead of const char*

make a version of Poohbear that has infinite memory cells in both directions

This requirement is impossible to test. My C solution was able to pass with a fixed 30,000 cells buffer. IMO, this requirement should be removed from the description, or at least marked as a "nice to have".

A quote from Kata Authoring Guidelines:

Do not use requirements which cannot be reliably tested, enforced, or expressed in terms of a kata

Ruby 3.0 should be enabled, see this to learn how to do it

Please organize the structure of test fixture as following (although it has been mentioned in the attached link, I'm repeated here again)

describe "<This message should describe the categories of test groups inside this block>" do
  it "<A short message describing this test group>" do
    expect(...).to eq(...) #Assertions
    #Or Test.assert_equals(user_response, reference_response)
  end
end

The test difficulty is not consistent across all languages: some languages permit brute-force solutions while others only permit the intended ones.

From PEP 8:

Function names should be lowercase, with words separated by underscores as necessary to improve readability.

AlphaNum_NumAlpha does not fit into any prevailing naming style. Please change this to alpha_num_num_alpha or similar.

( JS )

Test don't check against input modification.

( They're not vulnerable to input modification, they just don't bite your head off if you do. If brute force should be unallowed ( see below ), I think input modification should be unallowed. )

The point is that it should work with any precision (although I cannot test for any precision because the time would run out).

Nope, there are many ways to calculate pi, some are fast, some are slow
eg. http://numbers.computation.free.fr/Constants/TinyPrograms/tinycodes.html
Your solution is just too slow.
Also, your solution cannot work with any precision because it returns a float, which cannot represent too many digits.

Needs Random Tests.

(This kata is only of any difficulty or fun if you follow the rules. Don't be a cheater!)

Nope. It is YOU, the kata author, who are responsible for restricting solutions that should not be passed (according to the desc).