Print

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

[schismatise]

This was one of the bug fixes in 0.34.02:

- made dwarves with cancelled jobs stop walking to their destination

Great stuff.

However, i've been noticing for a while now that soldiers seem to exhibit a similar bug - i'll often find them wandering out into the field for no apparent reason, usually displaying "Soldier <cannot folder orders>" or some such as their current job. Quite annoying when it results in your legendary militia commander suiciding into an incoming siege. From what i've observed, they are usually heading to a location that you recently gave a Station order to, but have since cancelled or replaced with another order. When they get to the location, they may pause for a moment, and then re-assess their directive and head on to whatever job they decide upon.

Before you post "they do that because such and such", please bear the following in mind - it took me a very long time playing with military to resolve a million other issues that was causing similar problems - marksdwarves trying to pick up new bolts out in the middle of a siege, troops ditching their uniform in order to go pick up a new slightly better quality item that was just made, soldiers waiting indefinitely for a demonstration that didn't happen, etc etc - i watched my troops very carefully, over many forts. Being somewhat OCD and a bit of a perfectionist, i found myself compelled to at least identify the cause of my erratic military behaviour, even if i couldn't solve the issue completely, at least i could avoid it. To this date, the only issue that is still hanging around is the one mentioned above (incidently, if you need any military advice, i'm happy to offer my experience).

Now, i can't be 100% sure that what i'm experiencing is in fact soldiers with cancelled station orders still walking to their destination, and therefore related to the similar bug that was fixed recently, but i am fairly certain there is something going on with old station orders confusing my soldiers. I've made a habbit of always setting a new station order back at the fort entrance before cancelling my orders, which has seemed to help, but hasn't solved the issue completely.

My next plan of attack (if i can ever be bothered) is to run a test fort in which i never give my soldiers any orders at all, and watch them for several years to see if they ever wander around randomly - maybe they're heading to an enemy that was there earlier and isn't anymore, maybe they're getting confused by the minimum attendance being higher than the number of dwarves in the squad, who knows - but just incase anyone else has experienced a similar issue, i thought i'd share my findings with y'all, in the hopes of coming to some sort of conclusion.

Thanks for reading

[Yaotzin]

My soldier just dodged himself into a volcano thanks to this :p

I ordered them to station next to the volcano, because the siege spawned there. I realised almost immediately that was a bad idea and stationed them back somewhere safe. They didn't listen though.

I don't know if they just don't listen, or if they kind of queue up their station orders - I didn't cancel then restation. Shouldn't have to anyway even if that works, so I count it as a bug.

[schismatise]

Yep, that sounds like it matches my experience.

Cancelling an order or giving them a new order never actually takes effect immediately, come to think of it, but usually after 1-3 seconds. If your soldier notices an enemy before the second order takes effect, then they'll run after them, of course. But sometimes it seems like they decide to finish the first order before following the second one - almost like the orders are queued up.

Order queueing would explain this bug in some way - if a dwarf is sleeping, resting, or otherwise busy when the station order is given, he may decide to run off and follow the order long after it has been cancelled or completed. This would also explain why making your last station order at your fort entrance doesn't completely solve the issue - some dwarves suffering this bug may walk to your fort entrance for no apparent reason, but others worse afflicted will head out to the previous station order first.

[Darchitect]

This drives me CRAY-CRAY. I have seen dwarves run through a horde of goblins to get to a previously set station order. I thought, well, at least they'll stop and kill them on their way through. No. They won't, unless directly engaged by one of the goblins.

I'm interested in discovering a way to definitively clear out their queue. I wonder if burrowing them would work... The only other way is to disband the squad, which usually means they drop all their equipment where they are standing. Which means the second you put them in a new squad, they'll want to run out into a battlefield and pick up the most inconveniently placed sword EVER.

[schismatise]

Hmmm. Well i'm glad to mostly confirm my thoughts on the issue... i suppose more testing is needed then.

Perhaps cycling their alert to inactive and back again might do the trick? That way they won't drop the uniform. I'll give it a try shortly...

My basic method of avoiding problems is to regularly give them fresh station commands to the training barracks at the fort entrance, usually after sieges when i think things are safe. Once they've all obeyed a station order, and potentially cleared out any old orders, you shouldn't have problems till the next siege - and then you just need to remember to repeat the process after the siege when things are safe again. Of course, sometimes you think it's safe when it really, definitely is not...!

Little bit more info and i might make an entry on the bugreport system, i couldn't find the issue there when i looked earlier, though there certainly are plenty of proposed military bugs.

Edit: I don't believe you can burrow military dwarves, though if they are inactive they should obey civilian burrow restrictions.

[robertheinrich]

On a related note... I have noticed that my soldiers never cancel their orders upon fulfillment (killed target) or semi-fulfillment (target escaped from the map). They just keep standing near the corpse or map edge and get unhappy over time. Is that a bug? While I would understand that they keep watching out at the border in case the enemy comes back, but it canīt be right that they keep watching a corpse, can it?

[nenjin]

That's been happening for a while, ever since the new military system went in. They tend to treat kill orders like "stand at post orders" for some reason, even after their target is dead. Not a huge issue, which is probably why he's never taken the time to fix it.

[Linenoise02]

I've found it effective to issue new station orders after any battle.  Station the military to a safe spot in the fort, wait for all of them to arrive, let em sit there for a couple days, and then cancel orders.  It seems to work - the military will still wander over to their last station point, but since it's someplace safe, it doesn't get them killed.

[JackOSpades]

I had this problem in the previous version, It mostly went away after I changed their uniforms to "Replaces Clothing"
rather than the default of "worn over clothing." so it seems to be part of a larger problem with uniform miss-matches.

[nightwhips]

Posting to confirm this issue and to watch.

Have had two experiences updating station orders and watching dwarves run to the old station location and then move to the new one.

[Nan]

It seems to me that dwarves have some kind of "idle" routine which they use to decide where to hang out. One of the candidate locations they might decide to hang out at, seems to be a place they've recently been to / performed a job. I suspect that along with the cancellation being outright ignored, this is another factor which causes dwarves to go back to their old station location.

For example, in some of my forts I atom-smash all the dwarves clothing. I did this by putting everyone into squads, stationing them in the atomsmasher, giving them a naked uniform, and cancelling the station order. For some time, dwarves would wander back into the atomsmasher - not for any reason such as to get an abandoned waterskin [and they still had naked-time uniform or a replace-clothing uniform], but just to hang out in it for a while. I had to put a locked door on the atomsmasher so I could lock it while doing the smashing.

So it seems to be some kind of "I'm not really sure what to do right now, well, lets just go to the last place I've been to" type thing. Dwarves with a specific order don't do this, but military dwarves with no higher orders than to train, do.

[Loud Whispers]

Soldiers that are on duty to train with nowhere set for them to train at, they'll wander off to any corner of the map (usually a meeting area) and just idle.

[schismatise]

Spoiler (click to show/hide)

I had this problem in the previous version, It mostly went away after I changed their uniforms to "Replaces Clothing"
rather than the default of "worn over clothing." so it seems to be part of a larger problem with uniform miss-matches.

I start my fort with worn over clothing but shortly switch to replaces clothing. I haven't noticed it effect the issue at all (my dwarves still run off after old station orders either way), though for the record i play without clothing, and rarely assign more than 1 item per equipment slot (head, armor, hands, etc), so i don't get alot of uniform issues (or any at all, really). I do remember having similar issues in the past though, and i believe there are still plenty of bugs with military equipment, in particular layering.

Spoiler (click to show/hide)

Soldiers that are on duty to train with nowhere set for them to train at, they'll wander off to any corner of the map (usually a meeting area) and just idle.

This is true. I have also noticed that if you leave the default '10 minimum' training settings in a squad with less than 10, the frequency of finding soldiers standing around random places with "soldier <cannot follow order>" jobs seems to increase slightly. However, they don't seem to wander off - rather, they just get stuck for a moment or 2, and then sort themselves out.

Spoiler (click to show/hide)

It seems to me that dwarves have some kind of "idle" routine which they use to decide where to hang out. One of the candidate locations they might decide to hang out at, seems to be a place they've recently been to / performed a job. I suspect that along with the cancellation being outright ignored, this is another factor which causes dwarves to go back to their old station location.

For example, in some of my forts I atom-smash all the dwarves clothing. I did this by putting everyone into squads, stationing them in the atomsmasher, giving them a naked uniform, and cancelling the station order. For some time, dwarves would wander back into the atomsmasher - not for any reason such as to get an abandoned waterskin [and they still had naked-time uniform or a replace-clothing uniform], but just to hang out in it for a while. I had to put a locked door on the atomsmasher so I could lock it while doing the smashing.

So it seems to be some kind of "I'm not really sure what to do right now, well, lets just go to the last place I've been to" type thing. Dwarves with a specific order don't do this, but military dwarves with no higher orders than to train, do.

This makes sense, and seems to be the main issue at hand. The question at this point is what triggers it - does giving a new station order cause any inactive dwarves to first follow the previous order before following the new one? Does it trigger when a dwarf is busy eg. sleeping while you give a new station order? Does the monthly rotation, which seems to cause military dwarves to become inactive for a moment before reactivating for the next month (a whole other issue causing plenty of negative thoughts), trigger them to look for a location to "idle" while they are inactive? If you atom-smashed old clothing/uniforms, it might even be possible that they think the item is still there, and they're heading over to store or pick up equipment that doesn't exist (edit: probably a seperate issue).

I suppose more testing is needed. Thanks all for the feedback thus var, i'll probably stick it in a bug report later if someone hasn't beat me to it.