[v3.3] creeping coins bug

[Magnate]

Well, in this case you kinda made that "nyah nyah" rule yourself, when you told the game to hide all coins from you...

Possibly what we need is a 'not quite squelched' mode.

Say the squelched objects:
Are shown, but in a custom attr (so are obviously squelched)
Do not appear in any lists or menus.
Do not convert objects into piles- they only appear if on blank floor.
Are not interacted with in any way.
(I haven't played with squelch much, so I'm not sure to what extent this differs from what we have now)

So while the mimic would be mimicking whatever symbol squelched coins use, it would still not come out of nowhere.

This is basically exactly how squelch used to work in the old days, with David Blackston's purple dots. I never understood why they weren't implemented, as they seemed the ideal compromise between the clutter of not having squelch and this kind of problem related to squelched stuff being totally invisible.

Kaypy is right though: if you squelch coins, you do not get to complain about coin mimics surprising you. It's as simple as that.

@bulian: it is not correct to say that Angband has always been a game of potential perfect information. To be more precise, you can say that Angband has always been a game of *reliable* detection, in that it's never lied - but in older versions there was plenty of information detection didn't give you (like the exact nature of monsters outside LoS - all you got was symbol + colour).

On d_m's questions, it's interesting to note that I describe myself as a refiner/optimiser, exactly like bulian, yet my answers are different:

1. IMO infallibility is boring, and randomness is good. I don't think @ should ever have 100% chance of succeeding at anything, ever. There should be a 0.00001% chance of "you spill your [rations/oil/whatever] all over the floor!". (Note to my detractors: yes I am taking the p*** a little bit here.) On the specific issue of detection, I think detection should be made *less* reliable at lower levels, with only very high level casters/devices giving perfect info. I think this particularly applies to object detection - it should *not* be possible to be certain you have found every object in the dungeon without actually exploring every room. Part of what makes the game boring is the new level / detect / no tempting objects / find stairs cycle that happens in the late game.

2. I have the exact opposite view to bulian. It doesn't matter how dangerous the traps are if you can reliably detect and avoid them. That just adds a pointlessly irritating detection step into playing - better to rip out traps altogether. IMO trap detection should be less reliable, and traps should create interesting challenges but not instakills or game-wreckers (e.g. stat swaps).

3. Again, my view is opposite to bulian's: lurkers *are* one exception in hundreds of monsters - that's the whole point of having them. I would strongly oppose limiting them to certain terrain types.

4. Instadeaths are bad, and easy avoidance is also bad. So we should have harder avoidance (less reliable detection) and no instadeaths. Having said that, I support Timo's view that very rare very OOD monsters are a good thing and add spice to the game. So detection does need to be pretty reliable for very OOD monsters. In fact, I'd be interested to try making Detect Monsters *only* show OOD monsters.

5. IMO the current detection paradigm is tedious and detracts from the enjoyment of the game. I think we should go backwards (unusual for me!) and make less information available - dTraps doesn't show all the traps, dObj only shows certain objects (metal? maybe it could vary by race/realm/whatever - lots of possibilities here), and dMon shows only OOD monsters ...

[kaypy]

One thing I would argue about lurkers is that they should have a non-floor tile once they are discovered. Now that we do *proper* mimicking, there's no need for the interface screw.

Oh, and mimics of all sorts should go back into lurk mode if they are out of sight for long enough. (if they don't already- I haven't checked the code)

[relic]

Lurkers are already unique in that they don't have a symbol even when detected. Once they are detected, you still don't know where are unless significant use of the look command is used.

Tip: A quick way of finding a Lurker in room without using the look command is to use the target command. If there are no other monsters in the room it will immediately show the Lurker. I often use this method if I forgot where the Lurker is.

[Timo Pietilä]

One exception in hundreds of monsters seems wrong. Lurkers are already unique in that they don't have a symbol even when detected. Once they are detected, you still don't know where are unless significant use of the look command is used.

Lurker is not the only one with that capability. It has a more dangerous cousin.

[d_m]

EDIT: This is a super long post. Sorry about the lack of brevity and repetition--I was writing it last night, forgot to post it, then came back this morning and edited it.

1. Is infallibility something that makes the game more or less interesting?

This question may have muddied the waters. The thing I was getting at is that many (most) things in the game have limitations, exceptions, or failure chances. ESP fails to detect mindless monsters. Devices almost always have a failure chance, spells usually have one too (except high level mages/priests). Scrolls don't fail but are defeated by blindness/confusion and easily burned up.

You mentioned detection as being "infallible" which is what made me think of this, given that much of the game (IMO) is working around the fact that most other actions have interesting limitations/failure chances.

2. Would the game be better with (A) no more traps (B) as it is (C) traps which are less dangerous but not infallibly detectable with a spell?

My opinion is that (A) or (C) are both better than (B). That is, I don't like the "feel" of detecting for traps constantly, and I think it is a drag (especially for new players). It's exactly one of these situations where the "optimal strategy" for dealing with traps is also the least interesting--past a point pretty early on you can assume that you will never trigger traps without meaning to, unless you make a mistaken move or get careless/risky with trying to disarm traps.

If you know that there's a small chance you'll take damage from walking around (either from lurkers or from traps you couldn't have found) then it just means you'll try to avoid running while low on HP. Obviously if traps are incredibly lethal this will be a problem, but this is a balance issue.

If traps were always visible that would be another thing I could live with--it's basically the status quo without requiring the constant keypresses and book-keeping.

Finally, if trap detection was a status that you either had or didn't have (e.g. ring of trap detection, or some kind of search score at 100%) then traps could be obviously on sight and I'd also be fine with it.

3. Would it be OK for lurkers to remain as they are if they could only show up on a certain (non-standard) color of floor (assuming we had different terrain types for floors)?

I'm fine with the way they are. I was just wondering if giving a bit more tactical information about where a lurker might be was enough to make them less bad.

I do think it'd be nice to give Lurkers a symbol once they are discovered. Maybe X?

4. Is it worse to have monsters which easily cause instadeath but which can be infallibly avoided by experts, or to have monsters which can occasionally caues instadeaths (possibly only by working in tandem) but which are not infallibly avoided by experts? For beginners? Experts? Average players?

The point here is simply that Angband is a harsh game, where when you die you start over. Beginning players are expected to die a lot and have to start over, which is what it is. One of the things that I think is suboptimal about the current situation is that there are well-understand, reliable methods for detecting and dealing with pretty much any foe. The effect of that is that experts know exactly how likely a given monster is to instakill them (by evaluating speed, resistance holes, AC, etc) and also how to use LOS tricks, brands/slays, the AI, spells, items, etc to nullify those threats, deal with summons, kill or avoid the monster, etc. Because reliable detection and reliable escapes are both available to an expert, developers respond by creating monsters that can kill in one turn, to catch an expert off-guard or punish a single mistake.

New and even average players will almost certainly not know this kind of optimized play, and are therefore constantly punished by these monsters created to challenge experts. While I don't have a direct path there, I think that it would be possible to improve the situation by (slightly) increasing uncertainty while simultaneously toning down e.g. summons and breath damage. This would preserve (or even increase) the difficulty for experts while making things easier for beginners (the game would still be MUCH harder for a beginner than an expert).

5. Is the game's detection paradigm (repeatedly doing 3 different kinds of detection to obtain "total level coverage") an asset? A problem? Both? Neither?

I think everyone agrees that the current detection paradigm is a bit annoying. Eddie made the point awhile ago that you don't want to limit detection in a way which removes tactical choices, which I'm sympathetic to. That said, taking calculated risks is part of Angband also: people make the point that if you take enough risks with a 1% failure rate, eventually you will fail (possibly catastrophically). So a little bit of uncertainty can go a long way.

In my opinion trap detection is the least interesting kind of detection, and should be obviated/removed somehow (either by removing traps, or making searching for traps an "automatic action" based on search, or making traps always visible, or something else).

Object detection is useful and exciting but arguably a bit too powerful... right now I personally rely on it heavily to cherry-pick the best gear lying on the floor. Even if it were less powerful I think it'd be very useful. I think it could be possible to split the current behavior up among several spells that different classes get: for instance, maybe one class gets a generalized "detect all stuff" which doesn't show flavor, another gets "detailed detect weapons/armor", another gets "detailed detect potions/scrolls" or devices, or whatnot. Maybe rogues get the "detailed detect all" spell, possibly at a higher level.

Monster detection and ESP are probably the most important kind of detection, and the one that feels least like an imposition. I don't mind the fact that ESP has gaps, and also don't mind the new gaps that lurkers/mimics create. There is talk about creating ESP-like abilities for subsets of monsters, which I think will be really useful and exciting.

Right now priests/paladins/warriors make do with detect evil for a long time, and it works pretty well. I think other kinds of detection of different subsets of creatures (e.g. detect living, detect undead) could be good, as could a more directional detection (scrying) which perfectly detected a smaller area. I think that making Detection less infallible/perfect/powerful creates all kinds of interesting niches for other kinds of detection.

Anyway, I am not sure there is any kind of community consensus on these issues, just wanted to bring them up here.

[Nomad]

Finally, if trap detection was a status that you either had or didn't have (e.g. ring of trap detection, or some kind of search score at 100%) then traps could be obviously on sight and I'd also be fine with it.

How about having a search radius, like infravision? At radius 0, you only find traps when you step on them and/or search manually, at 1 you automatically find adjacent traps and secret doors, 2 you spot them from two squares away, etc. It would take the tedium out of trap detection, but also make it more difficult to plot your route through vaults to avoid them; your search radius would detect traps infallibly, but not until you got pretty close to them.

[d_m]

How about having a search radius, like infravision? At radius 0, you only find traps when you step on them and/or search manually, at 1 you automatically find adjacent traps and secret doors, 2 you spot them from two squares away, etc. It would take the tedium out of trap detection, but also make it more difficult to plot your route through vaults to avoid them; your search radius would detect traps infallibly, but not until you got pretty close to them.

Yes, I like that approach a lot.

I think one of the variants (Daj?) has something like that.

[bulian]

I appreciate the view points from Magnate and DM, as I think its interesting to hear the direction that the game is going from the people doing the work. I do appreciate the number of hours you guys are putting in.

Kaypy is right though: if you squelch coins, you do not get to complain about coin mimics surprising you. It's as simple as that.

I disagree with this cavalier attitude, but clearly I won't be able to change anyone's mind, so I won't be adding anything else to this thread.

For my own part, I will be commenting out creeping coins of the monster.txt file in any games of 3.3 I play. As comp 107 is an ironman game, I wouldn't be surprised to hear other comments about this. We'll see.

[Magnate]

I appreciate the view points from Magnate and DM, as I think its interesting to hear the direction that the game is going from the people doing the work. I do appreciate the number of hours you guys are putting in.

Thank you.

I disagree with this cavalier attitude, but clearly I won't be able to change anyone's mind, so I won't be adding anything else to this thread.

For my own part, I will be commenting out creeping coins of the monster.txt file in any games of 3.3 I play. As comp 107 is an ironman game, I wouldn't be surprised to hear other comments about this. We'll see.

I apologise if my attitude seemed cavalier to people playing - I am trying to use takkaria's distinction between gameplay and UI. Squelch is part of the UI, and my point is that issues caused by squelch should not cause people to complain about gameplay. This is doubly true when it's an aspect of the UI which is entirely optional. It seems perverse to me that you would comment out creeping coins in order to continue to squelch gold. But I am glad that we have a game which is sufficiently easily customisable that you can do what you want to do with it.

[buzzkill]

I think one of the variants (Daj?) has something like that.

No, DaJ has good ole' standard trap detection. Ewert apparently has implemented line-of-sight trap detection.

With the massive 3.2 changes I will have to manually recode a lot, but will be doing it soonish. Will do a zip file with changed files afterwards.

IIRC it is 4 square LoS detection with one in (range) chance of base chance. Base can be got over 100% eventually. Or get high dex and you won't trigger them. Searching is -2 speed and +20% base search, so at early levels puttinb it on in vaults etc is a good idea.

In my playtests it has worked nicely. I have once lost a GV to falling down a trapdoor due to inattention. :P Overall I like it. Oh and all trap detection is gone. And it finds secret doors too, which is cool.

[CunningGabe]

Here are my 2+ cents:

1. Is infallibility something that makes the game more or less interesting?

I'd say there are a lot of places where infallibility makes the game less interesting. However, I think it is important for certain actions to have infallible domains; e.g., scrolls are infallible as long as you are not blind or confused.

2. Would the game be better with (A) no more traps (B) as it is (C) traps which are less dangerous but not infallibly detectable with a spell?

I'd lean toward C, and I would prefer A to B. What I really want, though, is to nuke traps from orbit and have a completely different trap system where even if you could infallibly detect traps, they would have some tactical effect. For instance, I've suggested having rooms that have magical fields that affect anyone (monster or player) inside. A room could grant -10 to speed while inside, for example.

3. Would it be OK for lurkers to remain as they are if they could only show up on a certain (non-standard) color of floor (assuming we had different terrain types for floors)?

There are a few reasons I'm okay with lurkers as-is (which is to say, as I made them!) First, lurkers are not particularly dangerous. Second, most of the time when they could be dangerous, you can take preventative measures to mitigate this fact. In principle, you could fire an arrow down a hallway before you run down, to make sure that no lurker lies in wait.

That said, I wouldn't object to lurkers and possibly general mimics being detected by trap detection if we changed the trap detection system to one of these passive systems people are discussing.

4. Is it worse to have monsters which easily cause instadeath but which can be infallibly avoided by experts, or to have monsters which can occasionally caues instadeaths (possibly only by working in tandem) but which are not infallibly avoided by experts? For beginners? Experts? Average players?

I'd say the first alternative is worse. Ideally we'd have some sort of smooth progression where the player playing for the first time might be instakilled by any of a number of things, to the expert who would only very rarely be instakilled. I think the early part of that progression as it exists in Angband is okay. Early instadeaths come from fighting something much tougher than you, or failing to have an escape route, or not covering all your resistances. Later on, though, I think the lessons are a little less transparent. In order to challenge experts, I think we need a lot of tactical components we can mix and match -- not just powerful monsters, but also interesting terrain, room effects, etc. At deeper levels, you can mix more of these components or mix in more complicated components, so that you don't have to put in monsters that instakill newbies just to keep the experts challenged.

5. Is the game's detection paradigm (repeatedly doing 3 different kinds of detection to obtain "total level coverage") an asset? A problem? Both? Neither?

A problem. Detection should be less reliable, and total detection should be less necessary.


If someone were to code up one of the trap-detection systems discussed in this thread (where you passively detect traps in a pval-dependent radius, say), I'd be happy to modify it to detect mimics, lurkers, and creeping coins too.

[bulian]

Ok, this is separate bug related to creeping coins. Used -dObj and saw a pile. Unless I'm mistaken piles are never generated. Opened the door, see adamintine coins. Sure enough its creeping coins monster on top of another object, or in this case pile of money. Save file is attached.

[CunningGabe]

Ok, this is separate bug related to creeping coins. Used -dObj and saw a pile. Unless I'm mistaken piles are never generated. Opened the door, see adamintine coins. Sure enough its creeping coins monster on top of another object, or in this case pile of money. Save file is attached.

Is this in 3.3.0 or the nightly? This behavior should be fixed in the nightly -- which is to say that it should continue to look like a pile when you open the door.

[bulian]

Is this in 3.3.0 or the nightly? This behavior should be fixed in the nightly -- which is to say that it should continue to look like a pile when you open the door.

3.3.0. IIRC, upon opening the door the pile switched to the $ symbol; closing the door left it as the $ symbol.

I'm a little surprised if that is intended - having a pile is a dead giveaway that the square has a mimic.

[Max Stats]

...having a pile is a dead giveaway that the square has a mimic.

One solution is to disallow mimics to be created on a square with an item.