Nice Kata! However, Java tests (random tests) dont really test the solution properly, as there are 99,9% chance that the time chunk will end with seconds. This way, u can just use "...and second(s)", as there are almost never a test ending with days/hours (which needs to be solved through "...days and 3 hours").
I suggest adding some manually writen tests exactly for those cases, otherwise not fully-solved Katas are getting through.

I wish this used oxford comma :<

`Expected :67 days, 2 hours, 11 minutes and 1 seconds

Actual :67 days, 2 hours, 11 minutes and 1 second`

Exepected should be 1 second not I seconds when testing for clojure

This should be added to javascript sample and attempt tests so actual and expected are fully shown:

const chai = require("chai");
chai.config.truncateThreshold=0;

[C#]
Naming violation in the method name. It should be Pascal-cased.

C language: my code is running smoothly on the terminal, giving right results, but on kata, right on the second sample test I get the following error:

Test Crashed Caught unexpected signal: SIGSEGV (11). Invalid memory access.

Another interesting behaviour is that I passed 108 tests when I attempt, being 100/100 from "random tests" and 7/7 from "futher", and 1 sample.

I'm probably messing up with dynamic memory allocation. But since it works on terminal, I don't know what is it.

I am using C and experiencing segfault even though I don't encounter any when I run it on my local machine. I tried adding printf() in various lines to try to find where the invalid access is happening but printf doesn't show. Any one know how to debug it?

Although I pass all the randomTest, it returns a java.lang.StringIndexOutOfBoundsException: begin 0, end 5, length 0

Can anybody explain me why?

Add tests with convoluted external border, like this:

+----------+
|          |
|          |
|   +---+  |
|   |   |  |
|   |   +--+
|   +-+
|     | +--+
|   +-+ |  |
|   |   |  |
|   +---+  |
|          |
+----------+ 

Hey, I'd just like to know if such a shape is a valid input:

'\n'.join(['+----------+'],
          ['|          |'],
          ['|   +------+'],
          ['|          |'],
          ['+----------+'])

Reference solution in python fails with test :

+------------+               
|            |
|    +--+    |                  
|    |  |    |                  
|    |  |    |                 
|    +--+    |
|            |              
+------------+

I implemented an inside edge tracing algorithm, and I made a few corner test cases that my early implementations had issues with that did not have matching problems in the passing criteria for this kata. A couple examples:

1. "Dead Ends":
   A shape that includes a space that an edge tracer needs to completely turn around in:
   +---+
   | |
   | +-+
   +-+

Or if you want a "dead end" in all directions:
+-+
| |
+-+

2. Shapes where corners are right next to each other:
   +---+
   | |
   | ++
   | ++
   +-+

It would be cool to see a few of these corner cases put into the acceptance criteria to make sure everyone's alrogithms are handling them all.

Lua translation