[Beta 11.1, volatile] Crash when ARC triggered traps?

[R-26 Lightspeed]

Version : Beta11.1,
Seed : EfficiencyMatterTriangulator,
Scoresheet : https://cogmind-api.gridsagegames.com/scoresheets/T4kaK3eidZ7gTg6xw
Location : -4/Proximity Caves


A while back, during my successful volatile mode run, the game crashed.

I entered -4/Proximity Caves (I know it was that location, because it starts with being in line of sight of a lone sentry, which i distinctly remember having to flee from again after reloading.)
I wasn't doing anything special, but shortly after entering that level, the game crashed.

After reloading, i sensed an ARC approaching some derelict traps at about the same time (turn-wise) as when the game had crashed. Maybe the ARC dying to traps had something to do with the crash? The game didn't crash again when the arc died this time, so with the random element of explosions in volatile, it's difficult to say if it was actually related.

Crash log attached, which was also automatically uploaded back then.


Sorry for not properly reporting this until now!

[Kyzrati]

Thanks for the report, I'll look into it. May have been fixed already given the version is a bit old (and some related things have been messed with since), but I'll need to have time to go back to the older code and confirm what it was in more detail later on.

[Kyzrati]

Alrighty, had some time to explore this one and tracked it down to specifically an ARC running into a powerful trap in Volatile mode, one that has enough power to destroy it and cause an emergency deploy, which also happens to then randomly destroy a robot during deployment which then also causes an explosion strong enough to try to kill the ARC again (not too unlikely in Volatile mode since everything explodes...). This can only happen via ARCs that are not within earshot or visual range.

It's a complex cyclical bug which won't be fixed for now since I think it's currently only possible in that mode, in this specific type of scenario, and almost no one even plays that old event anymore so it's not really worth all the time it'll take to address unless it will also do something good for the regular game.

I'll just leave this one in the known issues section.

[R-26 Lightspeed]

Does it have to be a trap killing the ARC, just an explosion, or can it be anything, from a 0b10 combat bot IFF system glitch to an exploding machine?
What "randomly destroys a robot during deployment"? The trap, the emergency deployment, or any explosion occurring on the same turn?

While i'm wondering the exact reasons for why being in visual/audio range makes a difference, the reason i'm asking these questions is because... well...

Spoiler

Imagine a group of robots large enough that, if a chain reaction were triggered, would probably cause at least one megaexplosion. Let's say, 120 robots.
Now, imagine that some of them, let's say 15-20, were ARCs.
Now, because there's a lot of robots, most of the ARCs would probably not be within visual/audio range if you were to kill the first robot of the group from a safe distance.
What i'm saying is, depending on the exact causes for this crash, a certain endgame location could be a rather consistent way of causing it.

Even if the odds of it being relevant to my attempts at a combat volatile win are low since i'd prefer avoiding that location, i'd appreciate knowing whether i could theoretically cause the biggest explosion chain in Cogmind history or if i would need to make frequent backups because it would pretty much always result in a crash. Thank you!

[close]

[Kyzrati]

Doesn't have to be a trap, no, that was just how I was able to reliably recreate it (it's probably more likely when dying from an explosion or explosive trap). The ultimate cause is the volatile explosion of a dying deployed robot in turn managing to destroy the ARC again before that process was completed.

[Kyzrati]

Note about your "biggest explosion chain in Cogmind history": Someone did this once before in the regular game (not Volatile mode), and you can technically cause so many explosions it will slow to a crawl and eventually crash. It takes a while to set up, but they had hundreds and hundreds going at once

There was later a certain mechanical tweak which would indirectly affect the strategy they used, making it more difficult to achieve now, but theoretically might still be possible.

[R-26 Lightspeed]

I thought you might want to know that there's one crash log you received was from me, probably due to the exact same cause as the one from this topic.
(crash log attached, in case you'd want to figure out which crash log in your list is useless.)

Note about your "biggest explosion chain in Cogmind history": Someone did this once before in the regular game (not Volatile mode), and you can technically cause so many explosions it will slow to a crawl and eventually crash. It takes a while to set up, but they had hundreds and hundreds going at once

There was later a certain mechanical tweak which would indirectly affect the strategy they used, making it more difficult to achieve now, but theoretically might still be possible.

I can think of two ways for such explosive potential, but both ideas would actually cause relatively short explosion chains, if a chain means the largest number of explosions caused by other explosions consecutively (For example, an EM explosion chain reacting ten reactors would only cause a chain of two (EM explosion -> chain reaction), easily beaten by some array of fusion modulators (or the like) where one would explode the next which would explode the next and so on, leading to a larger chain).

Doesn't have to be a trap, no, that was just how I was able to reliably recreate it (it's probably more likely when dying from an explosion or explosive trap). The ultimate cause is the volatile explosion of a dying deployed robot in turn managing to destroy the ARC again before that process was completed.

I see, but why does the deployed robot die? Due to the intrinsic nature of emergency deployments, or due to something else?

[Kyzrati]

I see, but why does the deployed robot die? Due to the intrinsic nature of emergency deployments, or due to something else?

Due to the initial explosion. ARC exploding due to death and deploys immediately -> explosion immediately applied to the bots it deployed -> those bots explode because it's volatile mode -> those explosions immediately applied back to the ARC before it has actually finished dying (because all of this is happening during the deployment part of its death, and being applied immediately).

This wouldn't happen if they're nearby (since then explosions are not immediate), or outside volatile mode really (because you don't otherwise have explosions that are caused like this).