Serious performance degredation over time

[Serdra]

Over time, two players battling each other leads to serious performance degradation with no readily apparent cause (no memory leakage as far as I can tell). Here's the simplest code I found that reproduces the issue:

def minimum_repo():
    player1 = poke_env.RandomPlayer(max_concurrent_battles=1)
    player2 = poke_env.RandomPlayer(max_concurrent_battles=1)

    start_time = time.time()
    while (time.time() - start_time) < 5400:
        asyncio.run(player1.battle_against(player2, n_battles=250))
        player1.reset_battles()
        player2.reset_battles()
        print(f"Elapsed: {time.time() - start_time:.2f} seconds\n")

The elapsed time between iterations starts and remains constant for awhile before increasing dramatically, as shown here:

Other things of note:

• When decreasing the time and running the above python script several times in a row without restarting pokemon-showdown, there's no issue. This leads me to believe the issue is in poke-env, but I can't be sure of that.
• This occurs no matter what I set n_battles to. If a single call to battle_against is used, with n_battles=9999999, the slowdown still eventually occurs. Same with n_battles=1
• I've also tried with SimpleHeuristic and MaxDamage player, with the same result.

I'm running the most updated version of pokemon-showdown and poke-env, using the default config as mentioned in the setup documentation.

[cameronangliss]

This is a great diagnosis, thanks! I'd be very excited to see somebody figure this out, I've tried but I didn't quite reach this level of diagnosis so this is incredibly helpful!

[cameronangliss]

I believe this problem no longer exists in my fork of poke-env on my thesis-old branch. @Serdra I'm not sure if it was recent updates to the main poke-env or if it's just in my branch, I haven't had the time to look into it, but let us know if this isn't a problem anymore! If the problem still exists in the main poke-env but not in my branch, I think I know what the culprit is.