Food clusters

[matthewr6]

find major clusters

[an-OK-squirrel]

I don't really feel like this would be good to merge. For me, it toggles from whereever it's going to a random point with no food, which makes progress slow. I don't know why this is happening.

[matthewr6]

How often does that happen? I've noticed it, but it never seemed to happen much and that's never what killed it. Also, I think the mega food cluster implementation balances that behavior. I'd like it if someone else looked into that as well, but it may just be an issue of lag or the functions we didn't write, as the nonexistent clusters (which is what those random points are) always disappear in a few seconds. If needed, though, I can implement the food clusters into master from a different branch, which may or may not fix the problem.

[an-OK-squirrel]

I'd like for you guys to do the "using buffers for <100"

[matthewr6]

I'll try that tonight if I have time

[an-OK-squirrel]

Okay. I still can't get over 50

[an-OK-squirrel]

And I can't see what the robot is thinking (which I use a lot for testing)

[Znapi]

It loaded for me, but alternating where it is going every frame is a problem. If the direction to the food is the opposite of the random direction, it just ends up sitting there.

[an-OK-squirrel]

Exactly.

[Znapi]

The problem stops if you remove the else-if in lines 977 to 984. Something must be wrong with the condition you are testing, because it isn't doing what it is supposed to.

[matthewr6]

Remove that if/else for now, then. It's one of the only parts I haven't tested

[Znapi]

Should we turn back on enemy buffer rendering?

[matthewr6]

What do you mean?

[Znapi]

You have buffer rendering commented out in this branch

[Znapi]

Here: https://github.com/ScratchAgarioBots/agario-bot/blob/food-clusters/bot.user.js#L762 And some filtering function is commented out there too.

[an-OK-squirrel]

Okay

[Znapi]

Idk, maybe we should leave the filter function commented out. I left the bot running for a while an it's highest score was 105. It was doing better without it for me.nvm, this whole time it has been doing good with the filtering on.

[Znapi]

Should I push a commit and merge with master? It is working great for me right now(or it is just the server). I got to 230, and to 630 earlier because someone hit a virus and put themselves in a corner.

[an-OK-squirrel]

I'm gonna try one more time

[an-OK-squirrel]

Bot gets 30 points, other one almost always gets >50...

[Znapi]

I got to 230 again, and right now I'm at 182. What is the "other one"?

[an-OK-squirrel]

master

[Znapi]

Maybe I should try a different server. The one it is in right now is kinda empty.

[an-OK-squirrel]

OK

[an-OK-squirrel]

Ok, what happened?

[Znapi]

It got ~40 twice, then 204. Master is in a slightly more empty server and is at 265.

Aww, it just hit a virus.

[an-OK-squirrel]

See? Master is better.

[Znapi]

Testing this is so hard. In some servers it is constantly surrounded, and in others it has the whole level to itself. Master got 65, now it is at 305

[an-OK-squirrel]

Servers are always busy for me.

[Znapi]

Hit a virus at 484. Should I make a commit anyways, but not merge yet?

[an-OK-squirrel]

Sure, i guess. I just feel that it would be best to keep using buffers..

[Znapi]

I don't know what is wrong with the super clusters. Maybe it is prioritizing them too much over closer, minor clusters?

I'm going to go for now.

[an-OK-squirrel]

BAI

I'll merge after someone does https://github.com/ScratchAgarioBots/agario-bot/pull/49#issuecomment-118065034

[matthewr6]

Finished it. Thinking about taking out mega clusters or moving them to a lower priority, or even having them activate after a certain size (maybe 100?)

[an-OK-squirrel]

Still not a good as master right now.

[matthewr6]

So what's the difference? It's just as good as bufferless right now, if not better, and we have to keep moving even if it temporarily gets worse.

[an-OK-squirrel]

Maybe it's a problem with the superclusters

[Znapi]

How are we defining super clusters? Clusters of clusters?

[Znapi]

Part of our problem might be is that when it goes for a cluster/super cluster, it doesn't always pick up all of the food in it. It might pick up one, and then leave, making the long trip to get to the cluster pointless.

It also got to 272, then went off in a random direction and got eaten.

[an-OK-squirrel]

Remove superclusters

[matthewr6]

Huh, thought I did. Must not have pushed it up.

[Znapi]

I think you did. I am playing with it and I had to uncomment the superclusters.

[matthewr6]

Yeah, I did https://github.com/ScratchAgarioBots/agario-bot/blob/food-clusters/bot.user.js#L950

[Znapi]

We should make it so that when it goes towards a food cluster, it picks up all of the food in it by going all the way to the center, and when it goes to a super cluster, it should go to each minor cluster in it one at a time.

[matthewr6]

The thing is, once half a minor cluster disappears, the cluster itself is not a cluster anymore.

[Znapi]

That is the problem we need to fix. It needs to go all the way to the center, and ignore how the cluster changes. It needs to remember these things inbetween frames, so we should declare some variables outside of findDestination. Maybe one variable that stores the location of the next minor cluster it is going to, and then for major clusters, another variable that is an array of minor cluster locations in the next major cluster it is going to.

[matthewr6]

Maybe merge this first? :P I'll start a new thing tomorrow or Sunday, probably

sorry, I've been busier than I thought I'd be :L

[an-OK-squirrel]

Me too :L I don't see why we need to merge though.

[matthewr6]

To get it into master...

[an-OK-squirrel]

Then make a backup branch :P

[Znapi]

What does this have that is different from master? Everything is commented out.