Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

For !

Against !

[RAM]

Like I've said a while ago, tunneling should only happen IF the sieging force is decent enough...which is why I've suggested Siege Camps for example. 

I had an idea about camps independently, though I haven't read your thread yet I figure that they will get tired quickly, maybe after 3 or 4 tiles, and then go to a camp to rest if there is one, or if not then just go back home after doing their part...

Note: There should be more TAGs for individual creatures. Example:
[TUNNELING:SOFT] -> This creature Only digs tunnels through soft materials
[TUNNELING:HARD] -> This creature should be able to dig through rock even

A valuable suggestion that I would like to see implemented with any tunnelling, but it won't solve anything, if hard exists on anything some people will complain, and soft alone isn't going to be a challenge, many areas don't have ANY soft ground, and anyone can dig beneath the soft areas...


possibly something more like surveyors who carefully examine and record discovered traps that they see and have time to record and relay that information if their notebook makes it off the map...
But honestly, who would go to all that effort, you are better off having your opponent use old traps with second-rate mechanisms and weak weapons than forcing them to constantly update...
Spies may come in handy though, particularly for finding good destinations for tunnels...

I imagine that AI digging would be very difficult to impliment... If they don't dig wet or hot stone, then you could simply just fill your moat with water. I can see them strictly into dirt, to encourage the player to build walls that extend down to the bedrock though.

I doubt it, first approach would be to just draw a line between a point on the surface and a target in the fortress.
 Or have a single dig designation phase when a siege turns up, it uses the standard algorithm except that diggable material is considered to have a restricted movement value, maybe 20 tiles, if your entrance isn't a complete maze, and your fortress interior is a distance from open ground, then they won't path through any rock, if you really want to make it interesting, have dead creatures drop a hostile-only progressively restricted zone that will make enemies gradually less enthusiastic about pathing through an area where many past attackers have died. Of course, this may cause the game to stop for a few minutes while it paths through the entire map, but it would only happen once per siege...

You could have walls given a pathing resistance based upon the quality of the wall, based upon whoever smoothed/engraved/built it...

What I would like to see:

A siege arrives, pulling wagons. They attack in force, but between your traps and archer towers they take heavy casualties and withdraw. They return to their wagons, which are dismantled s to build a supply depot and some catapults. Soon your towers are broken, but you pulled your archers back and flooded the tunnels behind them, those paths will be blocked until crocodile-riding goblins with bags over their heads are implemented...
 They send some engineers in with some escorts. Many traps are disarmed, but some go undiscovered, and some are triggered. after losing about a quarter of their engineers they pull back, walls are built d around their camp while another assault is launched. The few remaining traps do good work, and you have underground battlements covering the approaches and a ballista bombarding them, soon they withdraw back to their camp, some desert, but most remain. Meanwhile you have not been idle, you have been digging a tunnel towards their camp. You undermine their depot, the building collapses with their food, much of which is destroyed, and the rest will start to rot if left unrecovered. Suddenly your tunnel is breached, you send your military in, but it is too late for your miner, who was waylaid from behind. A pitched battle erupts in the narrow tunnel. Your miners are sent to make a small room behind the battle when suddenly the tunnel widens, a goblin engineer loses an arm in the effort, but now your troops are fighting two to one. Your dwarves pull back to the room, where they fight three abreast, the goblins suffer massive casualties and withdraw. Injured and unprepared, you do not pursue. you seal up the tunnel behind your troops and wait, without food the goblins soon pack up and leave...

In time a team of goblin spies sneak into the fortress, some of them die, noisily, so the others try to run away. One is mauled by a dog, another runs into a guard, another is spotted by a hauler and is shot in the back from a battlement, but one escapes, and makes it back to their tower.

Soon another siege arrives, once again they attack, once again they are driven off and make camp. You see them start to tunnel, they seem to tire quickly, usually only digging two or three tiles before returning to their camp to rest. The tunnel is on a direct line to your troop's forward supply depot. You assemble your troops there and wait...
 The wall is breached in three places, goblins start trying to rush your troops through the narrow openings, Things seem to be going well, when you notice the rest of their force rushing through the entrance. Many have died on the traps, but many more are past them. It is too late to man the ballista, you rush your military to a choke-point. The few remaining at the breach are on almost equal terms, fighting bravely, while the bulk of your forces is nearby holding back the tide. You notice engineers in the main entrance, disarming traps and gleefully running off with the components. After a long and costly battle the goblins withdraw to their camp, while you set about sealing the breach, and tending to your wounded. With your army in no state to fight, and a now much reduced enemy force with full supplies, it will be a long time before you hear word from the outside world...

So... mostly I want them to only dig a few tiles before getting tired and taking a rest, either in a camp if they have one, or leaving the siege if they don't. And, obviously, I would love an easy way to have masons fill in an entire corridor without having to attend to each tile individually for fear of my masons getting trapped...

[NW_Kohaku]

Before I start, I have to mention that I have not read the entire thread. It's 22 pages long and I really have no desire to see if somewhere ten pages back someone brought up and discarded an idea similar to mine.

People are discussing the "swiss cheese" problem with tunneling invaders, and I do see this as a bit of an issue for an unprepared fort. However, if I was tasked to defend against a bunch of dwarves trying to tunnel into my underground fortress, I would simply prepare a proper moat. None of this one tile deep and wide nonsense that somehow manages to destroy pathing, but a proper deep moat that makes it impossible to dig through the top X levels without those miners flooding themselves with water or magma. The AI we talked about would avoid making caveins and not dig through warm or wet squares. Without that AI, I assume the invaders would die anyway, but it would make refilling the moat rather important. I also believe this means that aquifers would act as a perfect defense line as well. (After all, we know how much effort must be taken to properly pierce those and I doubt Toady will feel like making an AI able to penetrate them without cheating) To defend a fortress, one would simply surround the entire thing with liquids.

It is possible that this strategy would be ineffective against a group of amphibious or magma-surviving tunneling enemies, but at that point, there really wouldn't be much that you could actually do.

Basically, as RAM said, the tunneling should work as if it had some massive (50+?) pathing disincentive before working.

So you make attackers tunnel so that a 1-z moat will no longer work?  OK, they'll dig down 1 Z-level, and then ramp back up.  So, you propose that we make 3-Z deep moats?  OK, they'll dig down 3-Z levels and then ramp back up.  So you'll make a moat 10-Z deep?  They'll dig down 10-Z levels.  Unless you honestly want to simply channel straight down to the magma sea, then the depth of a moat makes no difference unless you limit the amount of distance that they will dig.

[rlbond86]

So you make attackers tunnel so that a 1-z moat will no longer work?  OK, they'll dig down 1 Z-level, and then ramp back up.  So, you propose that we make 3-Z deep moats?  OK, they'll dig down 3-Z levels and then ramp back up.  So you'll make a moat 10-Z deep?  They'll dig down 10-Z levels.  Unless you honestly want to simply channel straight down to the magma sea, then the depth of a moat makes no difference unless you limit the amount of distance that they will dig.

Even then, enemies should be able to construct ad-hoc bridges, too.

[thijser]

And how about using siege units instead. Units that can break trough maybe 1 tile of wal each 100 frames (just saying something balance needed)

[Solace]

if I was tasked to defend against a bunch of dwarves trying to tunnel into my underground fortress, I would simply prepare a proper moat. None of this one tile deep and wide nonsense that somehow manages to destroy pathing, but a proper deep moat that makes it impossible to dig through the top X levels without those miners flooding themselves with water or magma.

That's a pretty good idea, but... totally digging out all access to the outside world, top side and bottom, will make your entire fortress collapse. You might... if stairs count as supports... flood water or magma through your base and out through the same supports to make them impassable, but doing all that would take a very long time, like, megaconstruction fodder type time. Admittedly for a megaconstructon the coolness factor is high, but this isn't something you'd expect everyone to have to do by their second year.

[Richards]

How many people know that this thread is from 2008? And how many people know that there will be an init to disable tunneling? There shouldn't be any debate about this.

[NW_Kohaku]

How many people know that this thread is from 2008? And how many people know that there will be an init to disable tunneling? There shouldn't be any debate about this.

The debate is to find a way that would make tunneling even work in the first place (without causing the "swiss-cheese" problem that completely nullifies all static defenses and, for that matter, architecture)...

This is something that Toady wants, but hasn't found a way to properly impliment.  Suggesting a good way of doing this would, therefore, be a perfectly good use of the thread.

Furthermore, this thread was completely ressurected from the middle of the heap for the purposes of continuing the argument, but that is not something that should be looked down upon, and, in fact, should be encouraged, as there are far too many people that are posting three or four "new" suggestions in a single weekend.

[RAM]

Swiss cheese can be avoided by a pathing system, you select a point, then generate a single tunnelling path to that point, and all tunnellers follow that path.

Select a point: You could have a point defined by the previous siege, whenever an invader is damaged by a trap the furthest point that they can see in the direction that they are travelling is recorded, unless they are retreating. So they hopefully generate a path to a point of open ground past the traps.
 Or you could mark the last open ground an invader voluntarily moved to before it died, or any any place that an invader triggered a pressure plate or trap as a Low Traffic zone and have invaders avoid them more as more invaders die there.
 Or you could just take an average location of all workshops and path to the nearest open ground.
 Spies with high resistance to traps and stealth could be used, they would work like snatchers or thieves, maybe pathing to a high-value production task, or an office holder, and a high-skill civilian. If the sight their target, they turn around and try to escape, if they escape then the location of their target is marked as a tunnelling destination.
 Various overlays like the depot access view could be used to stop them getting into too much trouble...

Pathing: Include ground in you pathing for tunnellers, this will increase the CPU usage of the pathing so you keep limited numbers of these paths. Mark soil as, maybe, 20 tiles of movement, maybe higher, you need to balance out that tunnelling will probably produce a more direct route. Stone could be 60, maybe adjusted for the type, add a bonus for tunnelling resistant stone, maybe if it is smoothed, maybe if it is engraved, steel walls might be over a thousand, to compensate for the fact that the tiles you want them to dig through are high resistance stones. Make high-resistance materials take longer to dig through, so that if they do decide to path through them that there is still an advantage to high resistance. Generate an overlay of safe areas, this would be tricky for fortresses with above-ground defences. Maybe average out the deaths of active invaders and try to avoid it. You pick a safe spot near the target, generate a path from the target to the safe spot, and start digging...

For some cohesive ideas:
 Some units are marked as tunnellers, they use a special pathing script that includes solid ground. If you have a 200 tile-long maze filled with traps and ballista with a single floor-section of soil between it and the surface then the tunnellers will make a shorter route... Some invaders probably protect tunnellers until there are no more tunnellers left. If you have a 100 tile-long shaft going into the side of a mountain, then it is unlikely to be bypassed unless it was very unpleasant. You could add a penalty to pathing for invader deaths. An unfortunate side-effect would be that you would expand the shaft and the invaders would avoid the middle for no obvious reason. Also invaders would path around traps, goblins don't seem the types to memorise a trap because their great uncle was killed by it. You could produce a penalty over a large area to give more of a feel of a bad area than a bad item. Also, particularly old entrances would be completely ignored by invaders. You could have penalties decrease over time, but that would require a timestamp on each event, or you could do it arbitrarily... Maybe they would reset when new siege equipment became available. Also, of nobody survives then at most there should just be a general disease over the most obvious path...

 A spy arrives at the fort, it paths to a bedridden dwarf, it doesn't set off any traps and it manages to avoid detection, probably because it doesn't actually path next to anyone and just waits until there is an openning. It marks down the location of one of the beds in your hospital and sneaks off of the map, at which point that enemy has a destination for tunnelling. They send a siege with tunnellers. The siege paths as normal but stops at a safe location and builds fortifications, catapults, and covered food stockpiles. In the middle of the defences the tunnellers designate a path, avoiding open ground as that would probably just lead to a defensive passage. They slowly dig towards your hospital and then pour in and murder all your trainees in their beds. There is probably an init option to see where their tunnel is, but even if it is off, you probably get a warning when they are tunnelling within a couple of tiles of a dwarf...

[NW_Kohaku]

I might be wrong, but I believe that invaders, until they actually see a dwarf, will simply path towards a meeting zone.  Meaning wherever you set your dwarves and unclaimed, uncaged animals to meet is wherever the invaders are going to path to.

The problem with using dynamic pathing weighting is that the seigers don't actually re-path currently until they have their orders mixed up by, say, seeing a dwarf.  Currently, I believe the way that seigers work is that there is a single leader unit who paths towards the meeting place, and the rest just keep pathing onto the tiles that the leader previously occupied.  (Having seiger squads that move as a unit, especially involving sending cannon fodder up front so the leader units, always the most experienced and dangerous, don't fall to the very first cage trap you set out, would make far more sense.)

There's also some other problems with the way that things attack you, currently.  Once, I had a dragon attack when I wasn't really prepared to face the creature.  So I just called everyone in, and waited for it to cross the drawbridge into my entry hall, where it would run into a waiting cage trap... except instead, the dragon cleverly deconstructed the drawbridge while he was about to cross it, sending him down a z-level, and cutting off his only way to path to my dwarves.  Instead, he seemed to angrily knock over my still-exposed windmills as well, before just meandering around bored for a while before I suppose he got hungry, and wandered off.  Things like "Do not destroy bridges you were planning to walk on" seem like they should be part of the AI.

This doesn't even go into the fairly simple problem of most of these enemies that attack actually having only 1 skill rank in their chosen weapons (at least, as of 40d), with even the "champions" only having 8.

Basically, the seiger AI is rock-bottom stupid.

Anyway, the stuff RAM says makes generally good sense, if we are going to get the sort of love it would take Toady to give to the AI to make all that stuff, although I don't think spies are strictly that necessary.

The question I have, though, is how, exactly, "old" enterances are handled?  I mean, is it just based on the specific tiles used?  Do we then just gradually migrate the doom trap enterance up one floor every couple years?  It makes sense if retreating warriors would tell their superiors of where the traps that killed the rest of the army were hidden, but does sending a champion out to head off a seige just create a random spot in a field that seigers just avoid, even when they should know the danger is no longer there?

[Patchy]

I voted no for now. While I'm not totally against it and think it would be cool to see eventually, there is jus a few too many other things I'd like to see fixed, improved, added, or changed before this. Chief among them would be more fps increasing fixes, as I see the tunneling thing being murder on the fps.

[Shaostoul]

My opinion on it is... all in due time.

I think this discussion is a much smaller branch of a larger one.

Tunneling creature wise, I'm all for it, it'd add a new element to the game, if it was an init setting, that'd please everyone.

The siege bit, I think sieges need much more complexity and ability. A siege is a hard thing to hold off and there's a lot more than tunneling and scaling walls to it. I think as things are implemented, we'll see more features arise, at Toadies pace of course.

The main thing I want to see.. boats and flying fortresses, where you can actually move with that object.

[NW_Kohaku]

The siege bit, I think sieges need much more complexity and ability. A siege is a hard thing to hold off and there's a lot more than tunneling and scaling walls to it. I think as things are implemented, we'll see more features arise, at Toadies pace of course.

If we're talking about real-life historical seiges, then the reason seiges worked was because seigers would have signficantly more military force than the people they were attacking.  Otherwise, the defenders would simply sally forth, and drive the attackers away.  Real-life castles often held off invaders while outnumbered 10-to-1. 

Even then, seiges were not affairs of actually trying to breach the gates, but simply a waiting contest to see who would run out of supplies before the other side, and could potentially last a couple of years.

DF simply isn't equipped to do that sort of thing.  Champions can level whole armies single-handedly, meaning there is little reason to hide behind the ramparts, and fortresses can effectively stay within their walls indefinitely with no need for outside supplies, as their farms are safe underground, and their water supplies infinite, anyway.  Worse still, a waiting game just isn't what the combat-oriented DF players want, anyway.

We need reworking of seiges to make them less predictable and more challenging, yes, but real life is not what you want to model this on.

[Lord Shonus]

Sorry, NW, but that's not entirely correct. Historically, it was very rare for a town or castle to fall purely by being outlasted.  The bulk of sieges terminated in a storming of the fortifications, usually aided by treachery. The important element that's lacking in DF is proper morale handling. In RL, having an army camped outside the city gates was a tremendous burden, especially on the civilians, even if they were able to resupply and were in no direct danger of a storm. In DF, dwarves simply don't notice unless someone gets killed.

[Shaostoul]

I was referring to multiple ideas of sieging...
-poisoning of water supply (some people build wells that lead directly to a river/lake) If dwarfs are injured and given poisoned water they die.
-catapults/ballista/trebuchet maybe these could be made to take down certain thickness walls or hurl certain objects (explosives/creatures).
-aerial units could drop explosives/poisons from above (if the fort wasn't entirely underground and shielded)
-tunnelers...
-wall destroyers/bridge builders/ladder builders/etc.
-all kinds of magic (when implicated)

When I think of a siege, I think of the enemy doing what it can to get in, not some boring waiting game (which isn't quite applicable to DF seeing as caravans are the only thing that comes from the outside and they aren't even needed). It's DF, there is endless possibilities. I was merely just saying it's not limited to "traditional" siege ideas and tunnelers.

Also on the subject of tunnelers, you could entirely make it so anything of a certain size and smaller doesn't actually produce a dug square, (if possible) a half square could be represented.

[RAM]

Of course, sometimes not having enough rose gold bars from the caravan can get people killed...