BoyBlue
Level: Smitemaster
Rank Points: 547
Registered: 03-25-2006
IP: Logged

Guards normally avoid striking bombs with their swords, but sometimes they will strike a bomb when they step off oremites. I think this happens when they step off oremites in a direction other than the one they were facing. And it's not just bombs -- they may also strike other sworded monsters that they would not normally strike. Is this by design?

mrimer
Level: Legendary Smitemaster

Rank Points: 5411
Registered: 02-04-2003
IP: Logged

BoyBlue wrote:
Guards normally avoid striking bombs with their swords, but sometimes they will strike a bomb when they step off oremites. I think this happens when they step off oremites in a direction other than the one they were facing. And it's not just bombs -- they may also strike other sworded monsters that they would not normally strike. Is this by design?

Hmmm...This isn't the intended behavior -- just something we didn't think about. I'm undecided on whether to change it. One reason against changing it would be that it probably wouldn't appear obvious in many situations why the guard didn't move into the place that would have caused him to stab the bomb. Logically, it makes sense, but I'm not so sure it would be readily apparent by looking at the room.

____________________________
Gandalf? Yes... That's what they used to call me. Gandalf the Grey. That was my name.
I am Gandalf the White. And I come back to you now at the turn of the tide.

Pilchard VIII
Level: Smiter
Rank Points: 392
Registered: 01-21-2006
IP: Logged

That's a shame, I already had a room designed around this for KPVIIID3.

____________________________
The DROD Dragon

Jatopian
Level: Smitemaster
Rank Points: 1842
Registered: 07-31-2005
IP: Logged

Pilchard VIII wrote:
That's a shame, I already had a room designed around this for KPVIIID3.

Shame on you. Did Erik not warn us about designing holds around coding bugs? He certainly did - go read your help files again.

____________________________
DROD has some really great music.
Make your pressure plates 3.0 style!
DROD architecture idea generator

Tahnan
Level: Smitemaster

Rank Points: 2463
Registered: 11-14-2005
IP: Logged

Jatopian wrote:

Pilchard VIII wrote:
That's a shame, I already had a room designed around this for KPVIIID3.

Shame on you. Did Erik not warn us about designing holds around coding bugs? He certainly did - go read your help files again.

Yeah, but who knew this was a bug and not a feature? Banjooie and I discovered this one night--well, he discovered it, and I played around with it and speculated on things you could do with it. I thought you could have an incredibly cool room in which you're swordless and have to maneuver guards on and off oremite to get them to kill one another.

TFMurphy
Level: Smitemaster
Rank Points: 3118
Registered: 06-11-2007
IP: Logged

File: DROD 3.0.2 Oremite Patch.txt (17.5 KB)
Downloaded 47 times.
License: Public Domain

Okay, this topic seems to have that fun little yellow arrow on it. And it's with a part of the code I find relatively easy to understand! On the flipside, it hasn't been decided what to do about it. Ah well.

Still, practice is practice, and playing with AI tweaks can be fun. I've compiled a patch for the 3.0.2 source code that fixes three issues with regards to Guards, Stalwarts and Slayers when combined with Oremites:

#1. Swordsmen considered anything with oremites on to be safe to stab, even if they were standing on a non-oremite square next to it. This issue is already fixed in 3.1, but that's not available yet.

#2. Stalwarts and Guards stepping off of oremites still considered their sword to be facing the direction it was when they started the move. That's the same issue as described in this thread. Slayers never change direction when moving, so they're immune to this.

#3. All swordsmen considered killing moves to be top priority... even if the killing move involved "stabbing" from a square they couldn't possibly have a sword on.

I think I've managed to fix all three of these issues while leaving non-oremite behaviour intact. Of course, this still leaves the problem of which of these *should* be fixed. Personally, I'm in favour of anything that makes Stalwarts not backstab me and actually be a little more useful, but I'm not sure how much the Guard and Slayer changes will affect existing TCB holds.

Slayers are, by the way, still monumentally stupid defensively on oremites, but given that there's always been elements they haven't been able to defend against, I don't see a problem with it. I just gave them the basic updates to put them on equal footing with Guards (as in, Slayers now respect issues #1 and #3, and were never affected by issue #2 in the first place).

The patch is listed in text format after the explanations and reasons for all the various changes I made. Again, 3.0.2 base source code only - it very likely won't work in 3.1 since I'm sure Mike had his own way of fixing one of the issues here.

Anyways, regardless of whether any of this actually gets *used* or not, it's still just more practice for me, and if anyone is curious enough (and can compile the source code themselves) to test this against various holds, be my guest.

(Once again: Public Domain license, since I've yet to see confirmation I should be using otherwise. The patch is pretty useless without the original files anyways)

EDIT: Not changed anything in the uploaded patch, but was playing around a bit more. Decided I'd try fixing Slayers being unable to pick squares next to orbs that *don't* have oremites on. This worked. And then I found a rather funny Slayer weakness that's been in since JtRH, which is particularly suicidal of them: they don't test that squares are safe to stab when they're looking for routes to doors. I cheerfully watched a group of slayers kill each other trying to get to an orb, and after adding a couple of bombs in the right places, watched them blow themselves up.

It was funny enough and an interesting enough quirk that doesn't *quite* seem to have been brought up that I can quickly see in a forum search, so I thought I'd mention it in passing.

EDIT 2: Forgot to set wSwordMovement for Stalwarts and Guards when they're swordless and right next to their target (I've made it so that they immediately turn to face their target as if they were "bump"ing them). Updated the patch to reflect that. It shouldn't have broken anything since they're already swordless, but it's nice to be consistent.

I should specifically note, by the way, that the Oremite changes make Stalwarts and Guards follow their target across Oremites *far* faster than they used to, because they don't stop to turn every five seconds if they happen to be facing the wrong direction. In this patch, they realise now that turning on a swordless square is futile (unless they have nothing better to do) because any movement they make next will overwrite any careful positioning they've just done. Of course, this just makes them marginally more dangerous than Wubbas, but they're definitely not as handicapped as they were. Take that as you will.

---

[Last edited by TFMurphy at 07-30-2007 09:49 PM : Small update to patch to correctly set wSwordMovement]

mrimer
Level: Legendary Smitemaster

Rank Points: 5411
Registered: 02-04-2003
IP: Logged

TFMurphy wrote:
#1. Swordsmen considered anything with oremites on to be safe to stab, even if they were standing on a non-oremite square next to it. This issue is already fixed in 3.1, but that's not available yet.

I didn't fix this too well in build42. Your fix seems correct.

#2. Stalwarts and Guards stepping off of oremites still considered their sword to be facing the direction it was when they started the move. That's the same issue as described in this thread. Slayers never change direction when moving, so they're immune to this.

This is something we just overlooked, so I'm glad to have a fix for this now. Definitely makes everyone seem less dumb around oremites.

#3. All swordsmen considered killing moves to be top priority... even if the killing move involved "stabbing" from a square they couldn't possibly have a sword on.

Thanks for fixing this. I'm fine with all three of these issues being fixed. Otherwise, they'd never be fixed, and that would be a shame.

...
I found a rather funny Slayer weakness that's been in since JtRH, which is particularly suicidal of them: they don't test that squares are safe to stab when they're looking for routes to doors. I cheerfully watched a group of slayers kill each other trying to get to an orb, and after adding a couple of bombs in the right places, watched them blow themselves up.

It was funny enough and an interesting enough quirk that doesn't *quite* seem to have been brought up that I can quickly see in a forum search, so I thought I'd mention it in passing.

Oh, man...That's also an oversight. I've never seen it mentioned either. I'd like to fix this if possible. I wonder if it would break any rooms. Who knows...

I should specifically note, by the way, that the Oremite changes make Stalwarts and Guards follow their target across Oremites *far* faster than they used to, because they don't stop to turn every five seconds if they happen to be facing the wrong direction. In this patch, they realise now that turning on a swordless square is futile (unless they have nothing better to do) because any movement they make next will overwrite any careful positioning they've just done. Of course, this just makes them marginally more dangerous than Wubbas, but they're definitely not as handicapped as they were. Take that as you will.

I like this. Even less stupidity is good!

____________________________
Gandalf? Yes... That's what they used to call me. Gandalf the Grey. That was my name.
I am Gandalf the White. And I come back to you now at the turn of the tide.

Nuntar
Level: Smitemaster

Rank Points: 5280
Registered: 02-20-2007
IP: Logged

Hi. I'm bumping this because I wanted to ask for clarification on a point. At the moment, after the fixes described above, the behaviour of guards on oremites is:

(1) Most of the time, guards understand they are on oremites. (They don't waste time turning to face their target, and they reject moves off oremites that would strike bombs.)

(Clearly an intended consequence of the above fix.)

(2) As an exception to (1), guards will prioritise a move that would be a killing move if they were armed. For example, if a guard is immediately west of Beethro in an oremite field, facing east, and Beethro steps NW, the guard will move NW because he forgets he is on oremites.

(Weird, but can be observed in pretty much any room where a guard and the player can both be on an oremite field. Changing this now would break too much, and there's no real point to doing so.)

(3) As a special case of (2), guards will prioritise a move of this type even if it makes them strike a bomb. This is, as far as I know, the only case in which a guard can be "tricked" into striking a bomb.

This feels (especially given the above discussion) like it's definitely an unintended consequence that would have been fixed if it had been noticed at the time. So my question is, is it worth fixing now? It doesn't make sense as a behaviour (given the overall decision to make guards able to understand oremites); it's very hard to discover if you don't know about it; it's a special case within a special case so it presumably comes up extremely rarely.

I happen to know that a hold in Architecture uses the behaviour, so it feels like the right time to bring this up for discussion. It will be good to have a definite decision one way or the other, rather than risk releasing a room that might get broken later.

____________________________
50th Skywatcher

Dragon Fogel
Level: Smitemaster
Rank Points: 2760
Registered: 06-21-2014
IP: Logged

Nuntar wrote:
It doesn't make sense as a behaviour (given the overall decision to make guards able to understand oremites)

I would agree here if not for point 2.

(2) As an exception to (1), guards will prioritise a move that would be a killing move if they were armed. For example, if a guard is immediately west of Beethro in an oremite field, facing east, and Beethro steps NW, the guard will move NW because he forgets he is on oremites.

Given that this happens, if point 3 were changed, then the behavior would be:

-Guards account for being on oremites, unless they can make a killing move, unless that move would strike an undesirable target.

As it currently is, the behavior can be summed up as:

-Guards account for being on oremites, unless they can make a killing move.

In short, changing this would create an exception within an exception. Given that the exception already exists, I think it makes sense not to create a sub-exception.

Consider this situation. A guard is on oremites, facing southeast. Beethro is two tiles east of the guard. If the guard moved east, he would be able to kill Beethro; instead (if he can) he moves northeast, because that makes the kill with his current orientation. But the oremites make him face northeast instead, and Beethro survives.

This implies to me that guards are not thinking through the consequence of their moves when they try to make a kill. We could speculate on all sorts of lore reasons for this, but personally I find "since they're not thinking things through clearly, they can be tricked into detonating bombs" to be less strange than making them suddenly thoughtful in that situation.

For what it's worth, I remember making use of this a couple of times going through Twelve Days of Christmas, though I don't recall the exact rooms offhand. (Well, one was Grab Bag but I don't remember the exact pass.) It didn't seem particularly unintended when I did, though.

Kalin
Level: Master Delver

Rank Points: 264
Registered: 01-25-2016
IP: Logged

Oh, is that what's happening? There's a room halfway through DKH where guards will sometimes stab bombs and I couldn't figure out why. The architect clearly knew, since one of the surviving guards will respond by saying "You idiot!"

Am I correct in saying that if a guard's brained move results in stabbing a bomb after stepping off oremites, they won't look for a safe unbrained move?

____________________________
he/him
~170th Skywatcher

hyperme
Level: Smitemaster

Rank Points: 1428
Registered: 06-23-2006
IP: Logged

Given the "exception to an exception" thing, I don't think making additional changes to how guards step off of oremites is a good idea.

Since it might break rooms as well, that to me is a good reason to keep things as they are.

____________________________
[Insert witty comment here]
Qzvlkx?