Every test reuses the same input object, so modifying the input will break the fixed tests, which would be very confusing.

I've deep-frozen the configuration objects in a new version of this kata. Not sure if it's smart to mutate function arguments in the first place though :P

Also, transitions in the random tests can be modified.

They too are now deep-frozen.

The way I see it, there are two ways around this problem:

1. Instead of checking for bit-to-bit accuracy, calculate the difference between test and actual images, and check if it's below a certain threshold. Problem with this is that it might allow some actual bugs go through, or it might not let a valid, but complex image go through

2. Instead of using pre-rendered data for the test-cases, generate the test data within the tests, so that differences in canvas rendering will also be present in the test images.

Check out this page: https://browserleaks.com/canvas

As you can see, it will give you a uniqueness percentage, of how your browser draws a set of drawing instructions. The uniqueness is likely quite high too. What this means, is that your browser produces a different image from other browsers, so by definition, expecting pixel-perfect, bit-to-bit accuracy in this Kata is misguided.

I don't like this. My perfectly valid solution is getting rejected, because of floating point errors. My math and drawing procedures are solid, but because the test cases expect bit-to-bit accuracy (which can be impossible to achieve cross-browser/cross-OS with Canvas), my solution isn't not going through.

For example, take a look at my drawing vs. test case here: https://i.imgur.com/q0YkNSp.png
You might not even see the difference, yet there are sub-pixel inaccuracies, and I'm absolutely sure it's not my code that's causing those. It's inherent to Canvas API implementations.

All inputs for which conversion to integer would be nonsensical (as is the case with Infinity), should return zero.

If it helps; this kata can be completed with under 40 characters of code.

Y is most definitely a vowel where I come from. It is pronounced as a vowel in words such as happY, todaY, rhYthm, sYllable, manY, analYse and many others.

Since this is a JavaScript test and not a test of specialized knowledge of english phonetics, I think it should be at least mentioned in the task description, if not omitted from test cases.

8008 is the correct input for 80085e-1, according to the test case as well.