Print

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

[Bumber]

What's the difference between a "secret passage" and a door linked to a lever?

[Shonai_Dweller]

What's the difference between a "secret passage" and a door linked to a lever?

One requires the little used adventurer command to 'search area thoroughly'. The other is an amusing way to drop magma on people.

[KittyTac]

Then you have to make NPC's that are smart enough to either go around the wall (dig through it/go over/under it etc) or find the correct lever to open the passage and proceed through, tying into smart enemies taking alternative routes.

Why would you do any of that?

Because enemies never dig into walled-in forts.

[bloop_bleep]

Okay, some extra explanation on pathing and why this is infeasible.

The way DF pathing is optimized, is by splitting the map into a grid of chunks. There are two levels of pathing data -- connections between chunks, and connections within the chunks themselves. Whenever a unit wants to path to a given location, it first looks up its current chunk and the location's chunk and, using the high-level connection data, sees if there is a possible path between those two chunks. It then computes the precise details of its path by looking at the connections within the chunks in its path. Whenever a wall is built, a door is locked, or other such pathing changes are made, the game updates only the in-chunk connections of the given chunk (and perhaps the chunk's connections to adjacent chunks as well, if required), which doesn't affect any pathing that doesn't involve that chunk.

This, however, depends upon an assumption that if one creature can path from A to B, another creature can as well (usually; building destroyers are a whole 'nother story), so the game doesn't have to store multiple sets of pathing data. This assumption breaks down, however, when secret passages are implemented, since whether a given creature can path from A to B may be dependent on whether the creature knows about a secret passage. The game would have to store quite possibly hundreds of different pathing data sets for each creature, at which point it would probably be more efficient to just use basic pathing, which would still be abysmally slow.

This presents a very big problem. I am afraid secret passage won't make it into DF, at least not until Toady comes up with some absolutely genius pathing algorithm for it, which might take a very long time and might not be worth the effort.

[Shonai_Dweller]

Nothing's infeasible. Toady already said he doesn't mind destroying the game and remaking it from scratch if that's what it takes (don't think it's that drastic for this idea though). Secret doors are in the dev notes, part of the treasure hunter arc, complete with a note that it requires "path-finding optimization". It may be decades before it gets implemented, but it'll turn up eventually.

[bloop_bleep]

Nothing's infeasible. Toady already said he doesn't mind destroying the game and remaking it from scratch if that's what it takes (don't think it's that drastic for this idea though). Secret doors are in the dev notes, part of the treasure hunter arc, complete with a note that it requires "path-finding optimization". It may be decades before it gets implemented, but it'll turn up eventually.

Well OK, now you've got me thinking about a possible solution to this...

EDIT: OK, how about every connection between chunks is tied to a logical expression in terms of whether certain secret passages are known. The pathing algorithm would be the same as before, except that now, when the algorithm wants to check whether two chunks are connected, it would evaluate the logical expression by plugging in 1s for all known secret passages and 0s for all unknown secret passages. Still slower than a simple look up, but it could work.

[Fleeting Frames]

That seems like it'd result in standard advice of "don't make secret passages" along with "don't make pet-impassable doors" for FPS sake.

It's an improvement, though. Can maybe use some shortcuts where there's no passages on map (in which cases it returns to current fast implementation).

[Five chickens]

I just built a 'secret' passage leading from the king's bedroom to the corridor leading to the bedrooms of two visiting female nobles.  Coincidence?  I think not!

Seriously, I think secret passages and/or doors are a cool idea.  Or as an alternative, perhaps certain doors could require the use of a key.

[Shonai_Dweller]

I just built a 'secret' passage leading from the king's bedroom to the corridor leading to the bedrooms of two visiting female nobles.  Coincidence?  I think not!

Seriously, I think secret passages and/or doors are a cool idea.  Or as an alternative, perhaps certain doors could require the use of a key.

A key would be nice to see. No more "overseer says Lock!" instant locking doors. Would need to fix pet pathing issues first though. It seems like with hidden spider pits in kobold caves, we're a step closer to this than we were previously.

However, I played Adventurer recently and my home site was a reclaimed kobold cave. All the dorfs were happily roaming about their new home, while I kept triggering traps and falling down hidden holes. So, looks like the npc trap detection/triggering was skipped altogether there. So not really any closer to player-made fortress mode secret areas.