I see lots of submitted programs for the puzzle Roller Coaster which don’t even pass the last IDE test ; isn’t there a problem with the validator data (not big enough ?)
Most of the solutions are quite naive and don’t do any cycle detection and modular removing of those cycles ; at most there is an optimisation on keeping the gains / first group.
Very few pass the simple test:
1 1000000000 1 1
Might be a relevant read: https://www.codingame.com/forum/t/publish-only-100-solution/1303/2
tl;dr: Seems like the validator wasn’t big enough, should be fixed. But maybe it isn’t fixed after all 
OK, thanks ^^
So it seems the validator was fixed about 3 month ago to conform to the IDE test ; but I guess the data are still quite low for both IDE and validator ; a C near 32b signed int limit won’t pass for most of published solutions, even recent. For this puzzle to be interesting, and earn the “difficult” section, accepted solutions should be O(N), and not O©
I have published an O(N) solution in Clojure.
Takes 15ms to calculate last testcase, instead of ~3500ms for O©