This was fun, thank you!

Thanks! I like to write in functional style when possible and next in Python is very useful for that. However, after 2 years it took me a moment to understand what that accumulate does.

Agreed on moving the 4 into cycle, it saves a lot of multiplications (though s was meant to represent the sign, expressed with +1 or -1).

I prefer to use math.isclose instead of abs to check if the floating point numbers are close because I can quickly recognize what that code is doing, but your mileage may vary.

My solution in Python reaches the "rectangles have only common faces (version 2)" random test then times out. Ported to Javascript, it passes in approx. 9 seconds.

Nice kata despite the clunky description. I did not even try to brute-force the solution because I thought it would time out (my bad). I ended up writing a solution that works for more general cases.

This was fun, thank you! I am particularly happy with the way I achieved numerical stability in Python.

That's clever! It confirms that 55-character limit and return type integer could both be enforced.

I asked myself the same question, this is not an easy kata to rank. I would say 5-kyu or 4-kyu if the number of islands in the test was choosen so that all brute-force solutions fail, but any "textbook" implementation of the correct algorithm succeeds, regardless of language-dependant optimizations.

If the implementation must be optimized (i.e. choosing the right data structure) in consideration of the high "density" of the bridges, then definitely 4-kyu, but it is not clear to me if that's what you required.

A performance edition could be created (2-kyu maybe) where the number of bridges is so high that some upfront reduction is necessary (see my other comment hidden as spoiler) leveraging the 2D nature of the points representing the islands.

...and that's how you disregard PEP8 with flair 😀

Thanks for this kata. I have a couple of suggestions:

• The description of the kata needs some tidying up. It introduces several definitions that are hard to follow, also due to the flat formatting of the text.
• In the Python initial code it would be helpful to include a "pass" statement as a placeholder in each method. This way, it won't raise any errors, and users can start implementing their solution directly, method by method.

Finally made it, barely beating the clock! Thanks for this super kata.

I tested a few solutions provided by others and they all seem to be in the same order of magnitude as mine. I guess for the 5000 islands case even small optimizations matter.

Tried two or three modifications of that already, but not the right ones apparently! On the upside I'm learning more about computational geometry.

Thanks for getting back to me. I wanted to use numpy for matrix multiplication in one of my first attempts. I've tried several approaches that solve the problem but are too slow on 5000 islands, unless I use scipy. I wonder if I should replicate "that algorithm" in scipy (not saying more to avoid spoliers), but it looks a bit too challenging for a 4-kyu kata. I must be missing something obvious here 🙂