mrbermell

> Seems good and I think you already explained to me how this is an improvement. Can you remind me by adding a line or two to the PR? Description...

I'll wait with merging this until #234 is merged. Because the github action workflow broke and #234 fixes that.

@njustesen , can you take a look please :) It's annoying that all github workflow tests fail because one of the flags in the current pip command is deprecated.

I expect this test to pass. It doesn't. ```python def test_sweltering_heat(): game = get_game_coin_toss() for player_away, player_home in zip(game.state.home_team.players[:6], game.state.away_team.players[:6]): player_away.state.heated = True player_home.state.heated = True game.step(Action(ActionType.START_GAME)) game.step(Action(ActionType.HEADS)) game.step(Action(ActionType.RECEIVE)) game.step(Action(ActionType.SETUP_FORMATION_SPREAD))...

> I think renaming the function to make it public is a good idea but I'd prefer not to do it now. Let's wait for after Bot Bowl IV. Yes,...

> When and how would you use `get_illegal_action_warning(action) -> str:`? Is it better than just keeping the warnings and only raise then if `config.debug = True` I'll try to make...

I think helper functions is the way to go! Similar to the game creating functions in tests.util but with more arguments.

Must be a very subtle thing since both sides win the game. Here are some thoughts: - Shouldn't that script always generate the same result? Since the seed is specified...

That should then rule out the pathfinding and the scripted bot. One way to narrow the search is to do a frequency analysis of the outcomes.

Indeed. Maybe we there's a way to make replays use the forward model and still be less susceptible by tweaks in the framework. Saving the seed and actions are neat...