why have my solution failed?

[AverageWhale]

hello guys. I made my solution and it passed all tests including 80m number test but on the site its shows that my solution failed.

[Pindrought]

Just curious, did you upload multiple solutions?

It seems like yours should work? Then again mine says it failed even though I passed every unit test I tried. Wondering if I did something wrong or if the multiple uploads broke it.

To save people from downloading it, this was AverageWhale's code

Code: Select all

#include "Parameters.h"

namespace Solution
{
	Output Solve(const Input& input)
	{
		unsigned int input_size = input.size();

		int current_difference = 0;
		unsigned short max_difference = 0;

		unsigned short min = input[0];

		for (unsigned int i = 0; i < input_size; ++i)
		{
			current_difference = input[i] - min;
			if (current_difference > max_difference)
				max_difference = current_difference;

			if (input[i] < min)
				min = input[i];
		}

		if (max_difference == 0)
			return -1;

		return max_difference;
	}
}

[chili]

I'm basically planning on making a feature where you can request manual review, but only for patrons (mainly because i don't wanna spend all my free time manually re-evaluating tons of submissions)

You can be the first guinnea pig pin

[chili]

If you upload multiple, it's fine
The last one uploaded is the only one that counts

[Pindrought]

I'm basically planning on making a feature where you can request manual review, but only for patrons (mainly because i don't wanna spend all my free time manually re-evaluating tons of submissions)

You can be the first guinnea pig pin

I appreciate it, but my code is a bit convoluted so I wouldn't waste time on it.

I can just download your solution and generate random sets/compare results until mine fails to determine what it was I had wrong.

I am confused as to how AverageWhale's failed though. Seems really straight forward.

[chili]

yeah, it *could* be some weird bug with the test framework. Maybe i'll take a look and see the exact seed that causes the failure. This is the first wet run, and obviously if there are any problems we wanna get then ironed out asap

[albinopapa]

After looking through your "Ultimate" solution, yes from what I can tell, our solutions are similar even if our code is different. Still, 44% success, that sucks lol.

I'm sure I messed up somewhere, I got so brain dead by the time I submitted my last one, I just verified that it passed all the tests and submitted, never to think about it again.

[Slidy]

I also re-submitted and had no issues, so I doubt it's that.

As for AverageWhale's issue I suspect it might have to do with signed/unsigned mixing going on.

Would be nice if the system gave you the input set that your solution failed on.

[chili]

Don't feel too bad, my ultimate solution also had a silly bug in it that caused it to fail against one of the datasets from the forum thread. The problem was nothing to do with the algorithm itself, just the function i used to distribute workload among threads wasn't ensuring that each thread has a multiple of 8 input elements to work with So if I were competing on the leaderboard with the rest of you guys, and that dataset had been included in the test suite, I would have gotten 0 algos as well .

And yes, I also suspect signed/unsigned mixing bullshit to be the root cause.

[Pindrought]

Ah ran some test cases and found the culprit of my issues. XD

Had some old logic I forgot to fix for the smaller samples where single threading was being used. RIP