|Thread title||Replies||Last modified|
|A Few Suggestions||3||01:45, 29 April 2013|
I was looking through the code of version 1.2, and I noticed some things you might want to fix in the next version. Here they are:
- I don't think the code in onBulletHit is doing what you think it's doing.
whileloop in the radar code is unnecessary. Also, Double.POSITIVE_INFINITY is a byte cheaper than 1 / 0.
- Infinity Radar is sloppy enough even with setAdjustRadarForGunTurn. If I were you, I would get rid of that and replace it with setAdjustGunForRobotTurn.
Math.mincall in your Energy Management code isn't really necessary.
- Finally, you can assign a new value to
d0inline in your aiming code to save a byte.
That is some very good codesize saving advice. This nano was based off a relatively poor performing robot in the first place.
onBulletHit being wrong is possible, it was just getEnergy last time. However that looked like an error to me, since it was suppose to be detecting enemy energy drops, and I wanted to remove the possibility of
With my compiler, 1/0.0 has the same byte size as Double.POSITIVE_INFINITY.
As for the d0, note my comment "since I use this twice in the following, I am not sure which one is called first, if I figure it out I can shave another byte off". I assume its the most deeply nested one, but I could be wrong.
Basically, in the
onBulletHit event, you want to adjust the enemy energy to account for the damage from the bullet that just hit, so it doesn't register a false Energy Drop. When you use a (mostly) constant bullet power, like Sabreur does, you can pre-calculate and hardcode the bullet damage to save a lot of Codesize. However, if your bullet power is highly variable, you need to adjust by
BulletHitEvent.getEnergy() gets the enemy energy AFTER the bullet hits and its energy drops. So I shouldn't need to adjust by the bullet power. Sure I might miss a bullet fire, but at the time I was low on codesize.
But after all the reductions I can actually fit that in.