• All the information before the methods' descriptions is garbage
• The description is talking about mutability for no reason
• The methods' descriptions are using "Haskell notation" for some reason
• The description mentions "None/null/nil/undefined/etc type" while the only supported language is Python
• __str__, __len__, and __getitem__ should be implemented instead of to_string, size, and get
• The "existence" tests are useless
• Some equality tests are written using test.expect with a useless error message
• Solutions which set initial_value in reduce to 0 by default shouldn't pass
• The "[" and "]" ban can be worked around easily
• It is possible to use built-in deque
• There're already a few katas about linked lists, and implementing more methods doesn't make it more interesting or unique

I happen to like the approach you've taken. But it's still a duplicate, and I'm conflicted if your approach is sufficient excuse for that.

In this case, it's a duplicate of a 7 kyu kata. It might be better to steal a 5 or 4 kyu one for it. ( I'd like to see that done actually. It may go over better than this one. )

I implemented squareRoot as Math.sqrt. If you're going to have us implement umpteen components, and test all those components, you could have taken the time to disable that. ( Note that I am not a fan of disabling things. Mostly it means solvers are doing busywork. Here, we explicitly are. )

Also, if you're going to have us implement all those components, check that the composition works. Compose them yourself, and test that composition. Do not fully rely on the user solution.

You're reimplementing assert.approximately. Well, that fits right in with the kata as a whole. :P

• square - built in
• average - turn math function into a program
• abs - built in
• improve - turn math function into a program
• good-enough? and try - basically the same as this kata
• sqrt - built in

None of the individual tasks are meaningful/new, and throwing them all into 1 kata doesn't make it any good.

• Missing random tests
• I don't like the idea of reinventing the wheel (like sqrt, moreover square and abs!) => what's the interest?

https://www.codewars.com/kata/reviews/5e2401f1f6a4500001b667f2/groups/5e241da5ceed8c0001fe14d2
Only individual functions are tested, so that solution doesn't really do the whole thing and it passes the tests.

Haskell translation