If the memory-mapped IO interface was wrapped in a C interface, would that be fast enough?

It's a substantial performance improvement for heavily read/written files, as the OS doesn't have to copy a bunch of bytes. It just has to point the MMU towards the files

@Alion I feel bad for Highway now. Glider's highway banditry code is very good. The problem I have and wish I could solve...would be when the ants get involved. If I try to wait for them to be Flying Solo...they interfere with each other when being bandits. If I make the highway banditry code priority, then the entire glider transitions the MOMENT it sees the highway.

Solved that problem by discriminating on whether or not a given ant THINKS its part of the glider formation during the "standing in highway space" check that happens first. If it can reasonably assume its part of a glider, it aborts highway robbery and glides. Then on top of that I needed to fix some logic to make the robbery more robust during a 16 player game

Just a heads up - the update will feature a very slightly modified pattern (lime swapped with cyan). This was done as a last resort to mis-ID of Medusa as part of Highway (which often caused invalid transition from stage 1 to stage 2).

ie. you're just swapping those two with each other?

that's a trivial fix on my end for both Glider and Vampire, as both use arrays of color values and changing two of them around isn't that big of a deal

Yep.

Also, the code to detect "is this highway" is different in both Vampire and Glider. I rewrote the detection code from scratch (the "is this the center?" is essentially the same, just due to how I check for the view being mirrored)

But both ants do the overall check differently.

(actually had to fix a bug in Glider's detection code a minute ago that detected Windmill's main rail as Highway, causing a deadlock!)

Still can't believe how many problems Medusa used to cause. Just put the two in a tournament and observe a very sad Highway.

Well, maybe not very sad, but in my testing she could terminate Highway in ~15% games...

ha

current code running in Dave's controller is showing me that Vampire is not effective enough. Possible that the Highway attack code is too permissive and it's not going after SlM or Windmill enough. Glider does have a 13.6 (3rd place) after about 17-20 games

Highway is 5th or 6th (~11.9)

anyway, its 1am, where did the time go @..@

@Draco18s IKR? My 8-hour marathon from a couple of days ago was initially planned to last ~4 hours...

Since when does Explorer do that?

Now that Medusa is dealt with, Explorer decided to screw with my rail... You'll see the finally rail pattern when I post the update Draco, 'cause at this point I'm not even sure which pattern I can use to avoid the wrath of otherwise forever-irrelevant entries...

hmm this whole hexagonal non-100% certainty thing is gonna take quite some work to work around. At least there aren't going to be that many patterns possible to match at a given moment with remembering where you probably are

my idea is to have a stack of actions needed to be done in the memory

I'm planning how to play the challenge already but it's still in the sandbox without a controller :p

ooh Sliding Pairs would work wonderfully

@dzaima A lot of things would work wonderfully.

One of my ideas is a essentially super-charged whatever @eaglgenes101 is doing. Since you can transfer food between workers, you can just order 'em the right way and bam. Instant food transfer.

And falling apart is less of a problem since you can communicate with ants via memory, so nobody can disturb recovery processes.

@Alion ooh, you can do that?

In the spec ;) One of the things that was bothering me with this challenge.

Now that I think about it, I should state this more explicitly in the spec... It's in the bottom of the Output section.

"A laden worker trying to move onto an unladen friendly worker or a friendly queen will give its food to their target instead."

@Alion mhm, took me a while to find, though players would probably read the whole thing unlike me now :p

@dzaima Probably. Still, won't hurt to make this more explicit.

Super-charges Sliding Pairs as well - they can exchange food (well, aside from being able to track exactly where they are via memory...)

I'm hoping that the new spec will generally super-charge all existing entries, while also allowing for more advanced powerhouses to emerge.

a problem I'd see is that food would be gained way too quickly now

That will need some balancing, yes. I don't know just how powerful entries can get there, though...

@Alion My guess is way too powerful :p

That could actually pose another challenge. What if entries were forced to steal food from others, because there's just no food left anywhere anymore?

Nah, scratch that. Defense is too easy...

an LS type entry would get ~200 food out of the 5000

@dzaima Oh. So 8 LS-type entries and bam, 1/3 of the food gone. Damn.

Hmm, could one entry reasonably vacuum the map?

Ehh, it's possible under current rules, so I bet yes..

Hmm, unless there were tiny amounts of food on the map. That could be interesting...

I'm talking about, like... 200?

@Alion but then starting out good or bad would pretty much determine everything

No, that's not enough. LS-types would be winning... Hm. How am I supposed to write this, so there aren't just rampant vacuum-types everywhere. Is that even bad, though?

very horrible estimations lead me to a SlP type bot could get ~65k food if there were that much

Would each entry having lots of food be bad...?

interesting idea: big map, scattered starter food, and some random places on the map where food spawns often later in the game

making entries want to move to it

and maybe the generators move around or something

That sounds a lot like the food farming aspect of @eaglgenes101's termite challenge.

@Alion oh I have probably not read a portion of this chat then

Well, maybe not a lot like the food farming aspect.

@dzaima Here it is

But being forced to move base? I mean, sure, it would be doable, but... Hm. I don't know.

@Alion well you're not forced, you can keep doing what you're doing.

and interesting things like bordering the generators with ants could happen

@dzaima Depends on how much food they spew out... If they spew out tons, then whoopsie, move base or get rekt. If they spew out barely anything, then what's the point?

@dzaima Hm, would do you mean?

@Alion when the generator has been pinpointed you could border ants around it so others don't get it

@Alion you'd still have to find them (and trust them)

@dzaima Oh, that. And farm it long-distance as well.

Oh, how about this? Generators are common and they spew out food quite often, but worker ants can kill each other (details pending)...

@Alion that would work too

another question is how big should the generators be

So a strategic fight instead...

Similar to that semi-dead post already in sandbox, though...

@Alion ant wars?

@dzaima I think it was called that, yeah.

It's different all around, but the killing aspect would probably have to be similar.

so you'd try to not run into another ants base. It'd kill vampires too

@dzaima Well, depends on the attacking mechanics...

"An unladen worker trying to move onto an enemy worker will kill that worker and gain one food instead. If the enemy worker was laden, it will drop its piece of food on the cell it occupied before its death." How's that sound? Stationary workers will have an advantage, but after killing a worker, they will gain food (and become unable to attack as a result). Queens are immortal and cannot attack.

so laden workers can't kill?

Mhm.

@Alion what if the enemy cell has food on it already?

Oh, good question... Right, since laden workers can move onto food, they can also die on a piece of food...

Food can stack on a single cell, I guess? This will also allow the generators to be one cell in size.

But all of this would make the challenge entirely different. Most things from here wouldn't work well over there. It would basically be "who can control the most generators".

...and I think that's a downgrade, if anything. Surely there can't be more ways to control generators than to control a map...

And that would also encourage edit wars between the raiders and the raided...

Langton's ant sucking food from Highway because they got locked up by a chipped Glider wing...

Langton's ant ended that game with 1000 food :c?

@Alion I think that generators would definitely have to be bigger than a single cell. Maybe, keeping up with the same idea, the dropped food would appear somewhere nearby?

@dzaima Dropped by generators, you mean?

@Alion no, dropped by the enemy ant

@dzaima It would be pretty annoying to search around for a single piece of dropped food...

@Alion well you wouldn't search around for it (well you could), it just wouldn't be lost

@dzaima Mhm. Well, generally speaking, the whole concept of killing ants is... meh. I don't think we're going the right way, is all I'm trying to say.

@Alion true

At least it doesn't fit my vision of what the challenge is supposed to be.

As well as those generators. I feel like those aren't too in line with what I want this challenge to be. Maybe I could justify adding food over time, but on a global scale, randomly.

the reason I suggested generators is because I feel like end-game would get too boring as the map is quite small and food amount - quite limited

@dzaima That's why I said could justify adding food over time randomly. That might be better than just forcing a "hug generators" phase onto entries.

it makes the challenge more than just explore the most area, which is pretty well thought out on the current QotH

...and it would achieve the goal of making late-game more interesting, right?

how about the chance that a certain cell on the map can spawn food is based on some noise? So there are pockets of more and less food. The gap doesn't need to be big, just noticable for more advanced entries to be able to get just a bit more by being more strategic

@dzaima Hmm. But to hug the most generators, you would have to find the most generators, so you'd have to explore territory quickly as well.

@dzaima That's stepping into probability-theory, and I don't know whether I like that or not...

@Alion that's hilarious

A few more fixes to Glider later and it might be competitive with Highway, Windmill, and SlM again. Found an issue where Glider would attempt to rebuild inside (a half-erased) part of Highway and Worker 2 would say "oh, I'm on a highway, see ya!" and wander off.

And here I am, still trying to figure out all the things that are wrong with Highway... Caught some asymmetry bugs, fixed a couple of things, but it's still hard stuck on 4th place...

eh, I think imma just update the SlM post to SlM3, hoping it does well

@Alion One of the bugs I've seen is that the queen gets stuck between two Worker 3s

Happens very regularly.

Its probably due to my interference as a parasitic leech, but I've done what I can to avoid sticking a worker between the highway queen and her buddy

@Draco18s That's probably related to #13 on my TODO list (should really make a git, even if just for tracking issues). Basically, the Queen does some weird stuff with her type 3s... Although that is the first time I heard about getting stuck between 2 type 3s. Mind providing an example?

Like, what are the circumstances.

I don't have a seed you can use, 1) because this only happens with my updated Glider and 2) haven't seen it happen outside Dave's controller yet

Ah, I see. Though I'm fairly certain that all problems related to multiple type 3s are going to get fixed in one fell swoop.