Funny to read this three years later.
I did exactly the same thing, for exactly the same reason.

Is it because the first piece here isn't minimal?

Duplicate issue. Closing.

Fixed

the second one IS expected. The first one cannot be, because of the requirements:

You are given an ASCII diagram, containing minus signs "-", plus signs "+", vertical bars "|" and whitespaces " ". Your task is to write a function which will break the diagram in the minimal pieces it is made of.

re-ranked to 6 kyu

This has been fixed, you may retry the kata if you wish (as your solution doesn't work anymore).

I've made quite a few changes to the kata removing the "your directions should be the same as mine" part. After the testing process changes your solution turned out to be INcorrect, so you could try rewriting your code and passing the kata again if you want.

they just are random.

isn't the below an equally valid solution?

nope, because:

The shortest path is defined as the path with the shortest euclidiean distance going from one node to the next.

What you said would be valid counting the number of jumps only. But this way, you do not take in account the actual distance.

If the exercise wanted me to deal with an array of objects, it should have passed it as input instead of making me deal with it. If I'm going to deal with it, I'm going to make an intelligent data structure.

It's out of spec anyway, so you don't have to think about these cases (or any kind of input validation).

Issue marked as resolved

It is exactly the same.
First calc and sum and THEN round down.

Should work now.