First of all, happy to join the forum. I’ve been active trying to get some nice algos, but haven’t been active in the community before. I’m the guy behind the gamma, EMP^ERROR and KeKroepes algos.
Currently when an algorithm has an error you need to debug, it either is an obvious errror that always occurs, or a very specific one. The obvious errors can be debugged by uploading your algo and quickly sparring it before it is removed due to the crash. This is an ugly method, but works quite decently to debug an algo. The more specific crashes that only happen in very specific situations are more troublesome however. Especially when you want to develop algorithms with a decent level of complexity. I ended up making a separate algo that can quickly hardcode the exact behavior of the opponent when the algo crashes. Then sparring the hardcoded copy with my exact algorithm will recreate the situation and throw the error. This is a very tedious process however.
The feature request I have is that for every algorithm you post, the log of the FIRST error ever encountered is mailed to the author. That would save a lot of time.
We will be having some internal discussions this week about adding a potential feature to allow players to challenge others in the playground, which should address this as you would be able to rematch players that you crashed against.
More details about such a feature will come during next week.
Thanks for your response. Directly challenging others sounds awesome! However, re-challenging an algorithm you already played against would result in the exact same match right? Only if that would provide access to the log of my algorithm would any new information be added. I would also guess that the other author has to agree to do a rematch, and if only your algorithm crashes, there is little incentive for him to follow through with the challenge.
Anyway, I’ll wait for your announcement to see how the new challenging feature will be implemented.
Haha, I heard about that algorithm, but I didn’t make it.
I am the person behind the gamma_x, EMP^ERROR_Vx.x and KeKroepes_x.x algorithms.
The algorithm I wrote to hardcode other algorithms is meant to copy algorithms that give my algo a crash, not the algo it is playing against.
I can simply initialize per turn: new filters, destructors, encryptors, removal, pings, scramblers and EMPs.
I go manually through the turns, match the relevant info in the code bullet app, and paste them in the relevant spots in my code.
Really simple but gets the job done. If people are interested in this code I guess I can clean it up and share. It doesn’t have any strategic component anyway. It does make it really easy to program static strategies…