Fun kata. I went down the path of replicating the encoder first, then trying to build the inverse function. Suffice it to say the math got hairier than I'd expected and I abandoned that path and brute-forced it 'the other way'.

I was a little disappointed to see that a lot of the accepted solutions have hard limits on the length of strings they can decode, or they did not get the underlying alphabet exactly right (the '=' sign at the end isn't obvious, but it's there).

i still can't fix it!

Just found my diff error... I pulled a sneaky on myself.

Very nice. I was on my way down this path but got a bit lost in the math so I abandoned it. Should have stuck with it, very satisfying solution.

Fun kata. I went down the path of replicating the encoder first, then trying to build the inverse function. Suffice it to say the math got hairier than I'd expected and I abandoned that path and brute-forced it 'the other way'.

I was a little disappointed to see that a lot of the accepted solutions have hard limits on the length of strings they can decode, or they did not get the underlying alphabet exactly right (the '=' sign at the end isn't obvious, but it's there).

I got the same error when I miscalculated previous rank.

If a user ranked 1 completes an activity ranked -1 they will receive 1 progress.

Dude. That's a damn cool solution. I mean, if you did that in production on my team I'd take your bloody head off, but.... Holy hell that's cool.

If you are getting almost every test case correct, but somehow there are around 11 left with the error:

It seems like those are broken test cases, but they're not. No spoiler, but maybe re-check your diff calculation.