So, the engine has a few quirks that I’m sure most people have noticed. These include:
Commands are processed a tick later than you’d expect
Weird behaviour at apogee of jumps: hero pauses for a tick at the top of the jump
Weird behaviour during steal/radar: all movement gets frozen, even freefall movement, and resumed the next tick
Ability to stick to walls
Inability to affect vertical momentum in flight, so you can’t short jump. It’s probably a feature, but who knows.
There are likely other quirks - I know my pather discovered a way to achieve jump height 4, but it only tries it in the most useless situations.
So, on to the question. We’re less than two weeks from the deadline. Can we expect the engine to be effectively frozen until that time? I’m sure many people (or neural nets ) have poured hours into building bots that can work around the quirks of the engine, and it would be rather unfortunate if that all has to be re-done with a last minute engine release.
To offer an opposing view: can we please get an update to the engine rather soon so that we can proceed with our training getting more consistent, logical behaviour that matches the description provided in the rules? Personally, I have put off any committal actions like training neural networks because of promised imminent updates (more than a week ago now…), which I felt was justified because it affects all participants equally. I would certainly feel unfairly disadvantaged if there was an engine freeze, and that the quality and integrity of the competition would be reduced.
I don’t really mind what happens, but I’m sure I’m not the only participant slightly frustrated by the lack of communication on new releases or delays thereof.
) have poured hours into building bots that can work around the quirks of the engine, and it would be rather unfortunate if that all has to be re-done with a last minute engine release.