Print

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

[Makbeth]

I know, right?

That was so tactless and factless that I thought she might be a troll.  Looks like she isn't, but she does seem to be in the habit of making condescending claims that have no connection with reality.

[Caldfir]

Toady isn't avoiding a multithreaded approach based on principle.  It simply isn't clear that the potential performance gains from a naïve attempt at parallelism would win out against the inherent failures of parallel computing.  Namely, the memory accesses of pathfinding are dense, and there are potential gains with regards to caching which disappear once threading is implemented (unless you're really super clever about it).  My point here is that if toady were to throw together some slap-dash multithreading for pathfinding, it would quite possibly end up being slower than a single-threaded approach. 

Back on track here - the above discussion regarding algorithms for improved hauling seem sort of unnecessary - just scaling the z-axis by a factor of ~5 would probably make things work intuitively enough that players wouldn't be frustrated by confusing stockpile pathing. 

Also: wow, Toady is a machine with the bugs lately, no?

[Makbeth]

Also: wow, Toady is a machine with the bugs lately, no?

True.  And for some reason it seems to have spread to the utility coders as well.  Peterix released a DFhack build for .23 within five hours of the release of the game.

I wonder if it has something to do with the fact that the anniversary of 31.01 is four days away.  The dark magics released on that day must be lingering in the world still, and they are beginning to resonate and strengthen again as the anniversary draws near, causing mini fey moods in rapid succession for all who have exposed themselves to the code, or the echoes of it.

[Narmio]

Back on track here - the above discussion regarding algorithms for improved hauling seem sort of unnecessary - just scaling the z-axis by a factor of ~5 would probably make things work intuitively enough that players wouldn't be frustrated by confusing stockpile pathing.

Ah, but this is only true if your fortress is a traditional sprawling set of rooms on only several Z-levels! 

Do we want the dwarven pathfinding and job selection algorithms to make those kinds of assumptions about fortress structure? What happens if your fortress is cylindrical in shape, or even spherical?  What happens when instead of being based off a single central staircase, there are multiple possible up/down paths at regular intervals? What happens when, because of need to access caverns or magma or whatever, your fortress is two separate enclaves separated by a 30-level staircase?

In each of those (and probably many more) situations a static weighting factor will just confuse the distance function and could produce worse results tan the naive Euclidian system used now.

[kaenneth]

Im wondering what DF goes if the 'closest' item that it picked is blocked and inaccessable; does it do pathfinding, fail, pick another, pathfind.... until successful?

What might be better is not checking the entire list of objects just to select the closest one...

Another way to do the search is by following dwarf-accessable paths starting from the workshop, until an appropriate item is found, or all paths dead end. The bonus being that whatever path is found to the item, can also be used by the dwarf to get the item.

There would have to be a limit on the distance searched before falling back to the enumeration method, to put a cap on the search in case the nearest item is many hundreds of tiles away; or doesn't exist at all.

That also depends on if a tile 'knows' what items are in it; or if checking what items are in a tile would also require going through the entire list... and going through the list may also be required to see if there are any valid items at all; no point doing any path calculations in that case; but then you've gone through the list, and simply summing the differances in x/y/z could have been done while doing so...

Being able to assign a stockpile for a workshop to 'take from' would pretty much eliminate most of that work anyway, and would allow players to better control what gets produced.

As for multithreading, using multiple threads is easy; but doing it correctly, is very hard.

[[5th]Lupin]

I've posted this a couple times, but I still really want to see descriptor-genetics fixed. Something broke them a while ago and every time you gen a world now you get civilizations full of completely physically identical clones with the same hair, eyes, skin, broadness, etc. And while it's not a huge problem, it's a big immersion breaker for me as it hampers the "visualization" of the dwarves and prevents me from assigning special roles based on appearance.

[Chocolatemilkgod]

Fixing marksdwarves and wagons. Those are the two bugs I can think of.

[vrga]

Fixing marksdwarves and wagons. Those are the two bugs I can think of.

Marksdwarves arent fixed in 31.25?


(didnt get to play much yet)

[Makbeth]

Also, the tendency of military dwarves to stop fighting as soon as a their fifth toe, right foot gets bruised and seek the hospital, only since they're in the middle of a fight they just end up running back and forth, and become pincushions if there's any enemy archers around.  Had two proficient squads in full armor get decimated by two squads of goblins this way last night. 

It was the most pathetic thing I've ever seen in a game.

[Urist Imiknorris]

Fixing marksdwarves and wagons. Those are the two bugs I can think of.

Marksdwarves arent fixed in 31.25?


(didnt get to play much yet)

They tend to want to suicide-rush enemies.

[vrga]

Fixing marksdwarves and wagons. Those are the two bugs I can think of.

Marksdwarves arent fixed in 31.25?


(didnt get to play much yet)

They tend to want to suicide-rush enemies.

what about the training?

(suicide rushes are easily prevented as i do have to make defenses in such a way that the only way for them to attack is to shoot shit.)

[Willfor]

Fixing marksdwarves and wagons. Those are the two bugs I can think of.

Marksdwarves arent fixed in 31.25?


(didnt get to play much yet)

They tend to want to suicide-rush enemies.

Spoiler: Image (click to show/hide)

[Makbeth]

I could live very happily if I never saw the status "Soldier (Cannot follow orders)" next to my marksdwarves ever again.

One has to wonder why he called the military arc finished when there are so many problems with ranged soldiers.  That's supposed to be EA's thing, not Toady's.

[Urist Imiknorris]

He hasn't actually started the military arc. That comes after the caravan arc.

[Makbeth]

I thought 31.x was the military arc.  At least, the first few releases of it.  That's where the new military screen, schedules, alerts, etc. came from.