[MOD 0.13.17+] Rampant

[Veden]

I love the idea of this mod. But unfortunately it makes the game easier not harder as it is. If you walk away from security and get near alien bases - it actually much easier as almost every time you attack aliens the all flee in terror, and come back after a while, making you kill of a few at a time.

What was your evolution at? Also do you play with any other enemy mods or warfare mods?

[EldVarg]

Could you make a trigger to destroying trees (after some initial time) that will trigger an attack?

Is there some progress making this mod harder?


Edit: Well my evolution was 0 to 0.4 when testing it, and it was always very very easy. Was only playing with Yuoki mods.

[Veden]

Could you make a trigger to destroying trees (after some initial time) that will trigger an attack?

Something similar to Natural Evolution?

Is there some progress making this mod harder?

Yes, I have made some progress in making it harder. Should have something out today.

Edit: Well my evolution was 0 to 0.4 when testing it, and it was always very very easy. Was only playing with Yuoki mods.

I usually play with at least bob's enemies and natural evolution enemies, which causes spawners to have more hp and biters to be stronger.
I will see what I can do for just standalone.

[Xuerian]

Thanks for sharing this mod, it makes aliens much more interesting to fight.

[EmielRegis]

I dont know if that interesting you but found some weird alien behaviors when using Robot Army mod. Alens tend to not detect robots in proximity of their base, other bases wont send reinforcements and also often attacked bases sends patrols that ignore just attacking droid squad and run gaben knows where. I assuming most your scripts require player to work. In result sending droid army without going myself makes clearing aliens very easy.

[Veden]

I dont know if that interesting you but found some weird alien behaviors when using Robot Army mod. Alens tend to not detect robots in proximity of their base, other bases wont send reinforcements and also often attacked bases sends patrols that ignore just attacking droid squad and run gaben knows where. I assuming most your scripts require player to work. In result sending droid army without going myself makes clearing aliens very easy.

You are correct in the the player is the current trigger. This restriction is something that I would like to lift in a future version.

[Demosthenex]

I just completed a map with NEE and Bob's enemies. I'm looking for a new challenge! I'm curious about Rampant, especially because it says it optimizing pathing.

NEE would tank our UPS anytime an attack started, whether it was from trees, killing hives and retaliation, or worms. The quantity of bugs and pathfinding would make the game unplayable. We'd drop the game speed to 30% in order to continue.

Would Rampant be an alternative that is UPS friendly while being harder?

[Veden]

Have you tried removing NEE and adding Rampant to see how it works?

If the biggest bottleneck is in fact the pathing then the groups generated by Rampant will be easier on the UPS.

**Note not all groups are generated just by Rampant, the vanilla AI can't be turned all the way off. Fixed in 0.15 though**

Playing single player I usually see an update time of roughly 50-80 ms with spikes of up to 120ms when combat is in full swing due to the retreat mechanic.
In multiplayer there is a small overhead associated with every player.

[Demosthenex]

Have you tried removing NEE and adding Rampant to see how it works?

If the biggest bottleneck is in fact the pathing then the groups generated by Rampant will be easier on the UPS.

I think that pathing is the problem. We just completed the game (viewtopic.php?f=8&t=39876) and so I'm planning for a new one.

[Veden]

What do your path finder numbers come up as in the debug overlay ingame?

[Veden]

the other thing may be the desertification on enemy death from NEE, which is configurable.

[Demosthenex]

What do your path finder numbers come up as in the debug overlay ingame?

125+ during attacks for Entity Update. I'm not clear on the other figures.

the other thing may be the desertification on enemy death from NEE, which is configurable.

I thought that just changed the biome to make pollution spread. I wouldn't expect there to be any persistent objects in the desert to consume CPU time.

[Veden]

Veden wrote:
the other thing may be the desertification on enemy death from NEE, which is configurable.


I thought that just changed the biome to make pollution spread. I wouldn't expect there to be any persistent objects in the desert to consume CPU time.

the scorched earth setting causes the ground under an enemy when they die to turn closer to desert. This tile updating can be very expensive when lots of enemies are being killed.

[Demosthenex]

the scorched earth setting causes the ground under an enemy when they die to turn closer to desert. This tile updating can be very expensive when lots of enemies are being killed.

It persists while they are angry. Given much of the area is already desert at walls, would that add up? I've also seen our UPS tank before they are being killed. It looks like when they plan a rush to attack the wall, the UPS goes down. I became the "canary" that indicated when an attack was coming because I couldn't move anymore and my jumps increased dramatically.

[Veden]

the tiles would not stack.

Building a rocket silo will also cause waves of biters. Which I believe would cause some pathing slow down, as it just grabs everything in a large radius and attacks a silo.
The waves happen at certain intervals and target silos.

that code exists in the expansion mod

[Betep3akata]

Hi, Veden!

Thanks for your awesome work on the field of making Factorio not only a production and an automatization challenge but also a survival.

What about a new force on the AI control?
For aimed diversions for example.

https://mods.factorio.com/mods/Betep3ak ... -the-wisps
I can provide the squad creation interfaces if needed. At the moment, the Will-o'-the-wisps mod may be quite heavy by CPU consumption but work in progress.

Thanks,
Alex

[Veden]

I like the idea of something that has a different mechanic to combat.

I haven't gotten a chance to play around with your mod yet, but it is on the todo list.

I have briefly glanced through the code and noticed you making a local reference in targeting.lua:16 to the random function, which will cause desyncs in multiplayer. The desync has something to do with the way factorio handles the random function in the global environment.

[Betep3akata]

Thanks! Very valuable remark!

I naively believed that math.random (without generator init) will produce values by a deterministic way for all hosts.
What do you think will be better and safer? Will the SomeCustomDetermeisticRandom(game.tick) an appropriate solution? Or even math.randomseed(game.tick) ?

Edit: Or just the local reference to the global math.random is a problem?

Thanks,
Alex

P.S.: it is strange that we didn't felt any desyncs on the headless server for more than 50 hours...

[daniel34]

I have briefly glanced through the code and noticed you making a local reference in targeting.lua:16 to the random function, which will cause desyncs in multiplayer. The desync has something to do with the way factorio handles the random function in the global environment.

I naively believed that math.random (without generator init) will produce values by a deterministic way for all hosts.
What do you think will be better and safer? Will the SomeCustomDetermeisticRandom(game.tick) an appropriate solution? Or even math.randomseed(game.tick) ?

Using math.random() in multiplayer doesn't cause desyncs. It is based on the map seed and deterministically written by the devs to always give the same result on all clients/hosts.

viewtopic.php?f=25&t=12451#p83708

math.random uses the determistic Factorio random based off the save game. That means you can use math.random all you want and it won't break the game or cause desyncs.

[Veden]

If it is working fine then disregard my comment, I remember having desyncs due to a local reference to the math.random function which caused the random number generators to have different generated numbers on different hosts.

A global reference math.random is perfectly fine in everything, I'm only talking about the local reference to the global function.

Again if everything is working for you then forget I mentioned anything.