cool

fixed in C#, Julia, JS

fixed C++

Hey! I deleted the fork, figured I am not sufficient at this right now. Thanks for the comment!!

Hello. Would you check the latest fork? I am currently trying to figure out how to properly do this. So, I would really appreciate your suggestions!

I will fork it in a couple of hours!

• Why is the solution defined in the class of the tests???? that's not testing the user's code.
• anything close to the ref solution must be private.

I'll reject this one because of the problem with the test holding the solution. Too risky to keep that around. You can either create a new one or fork this one to make the required modifications.

Very polite explanation tx

Yep, same thing is happening with me. I've passed 208 cases and failed because of 1, and the reason is the maximum stack calling issue.

I never insist on biased random tests; it can be a fixed test, which could also be a nice small case.

Biased random generation could start with generating three numbers x < y < z that all have gcd /= 1 pairwise ( primes is simplest ), generating [ x, y, x*z, x*y*z ], then possibly adding more numbers as long as they don't include z.

'think so.

Is it better now?

Looks to me like the tables' headers are a bit misleading:

• "product of" describes how one value in one row is built
• the left column actually shows the value for that product...
• ... while the header is closure({2,3}), which is actually the whole possible values (hence rows) together.

So, in my comprehension, closure({2,3}) should be the title of the whole table, and not the header of one column. Same kind of problems goes with the second table (but not exactly the same, since the information is organized a bit differently)

edit: I'd even give the products first, then the values in a second column. But that's a matter of preferences, at this point.

very nice

spoiler flag, please...