Ramblings on latest RC

[jens]

So, I took a poke at the latest today (june 26 - 9367950). Doing this in one post, since it's mostly rambling, and too many different things to make posts about...

###

When starting new player, you can leave the name field blank, and get a character without name. This character can then be saved and get's a save file called PLAYER.

#
Monster drains your mana.
It notes that you don't have any mana.
Then drain you again.
And again note that you have no mana.
And keep on... First when this appeared I assumed it was to show that the monsters could learn, but now I'm not so sure they are learning...

#
When identifying something in inventory, would it not be more helpful if the reordering of items happened first? I know, not strictly logical, but more helpful:
- read scroll of inventory on wand in slot q, output from game:
In your pack: Wand ... (q)
You have 3 scrolls of Identify (j)
You reorder some items in your pack.

Now the wand has moved to slot (o), but I need to check that if I want to find out. Would be helpful if the message game the relevant information from the start...

#
DSM: All enchanted items should be considered magical, that the base is an enchantment of +10 does mean that +10 is average, but it is also magical, and it would be more consistent with player intuition to say that it is magical.

#
Too many DSM, and too early as well, not saying that they are too good at that level, but just feels like it's too early. Maybe I'm just used to them taking some time to show up...

#
Dropping too many egos and artifacts. I suggest we go and tweak that number somewhere that sets the chance for egos to be created...

#
Halucination lasts for too long. Even if we want for it too be a really long and annoying effect we could cut it in half and it would still be.

#
Wand/staff of slow monster still too good. Slowed uniques twice on second attempt, 6 on first attempt (OTH never succeded a second time) (playing H-Troll Warrior). And then usually tough uniques were push overs. I always had time to kill the monsters I had slowed before they sped up. My suggestion is to make it just a bit harder to fail, but shorten the time a lot, so they are only slowed like 4 turns or so.

#
I really like the new spikes and doors. I think it might really bring out new plays, at least in the early game. But, stuck doors are too hard to bash down. My H-Troll Warrior doesn't really have a problem, but earlier when I was playing a mage he could never open a stuck door. I wasn't really testing it, and he only encountered 2 doors.

[Derakon]

Regarding Slow Monster: harder to fail, or harder to succeed? I.e. should the wands be more reliable or less?

[Magnate]

When starting new player, you can leave the name field blank, and get a character without name. This character can then be saved and get's a save file called PLAYER.

Not on unix it doesn't. What would you prefer it to be called? Or do you want it to prompt for a charname on saving?

Monster drains your mana.
It notes that you don't have any mana.
Then drain you again.
And again note that you have no mana.
And keep on... First when this appeared I assumed it was to show that the monsters could learn, but now I'm not so sure they are learning...

Ok, this could be a bug - are you playing with ai_smart_learn on? And are you referring to non-STUPID monsters (i.e. not jellies / mushroom patches)? If yes to both, please post a savefile which shows this happening and I'll try to debug it.

DSM: All enchanted items should be considered magical, that the base is an enchantment of +10 does mean that +10 is average, but it is also magical, and it would be more consistent with player intuition to say that it is magical.

Are you talking about pseudo-ID inscriptions? I rather hope that Eddie's next generation ID will get rid of pseudo altogether, and all these horrible anomalies will go away.

Too many DSM, and too early as well, not saying that they are too good at that level, but just feels like it's too early. Maybe I'm just used to them taking some time to show up...
#
Dropping too many egos and artifacts. I suggest we go and tweak that number somewhere that sets the chance for egos to be created...

Increasing rarity of DSM is easy, though I'm keen to hear from other people if they are agree that DSMs are too common (now that there are fewer egos). Reducing overall numbers of egos and artifacts is trickier. I'd prefer to wait until 3.4, because rewriting the artifact generation to use the "special artifact" route is going to need a whole lot of rebalancing (thanks to Derakon for the baseline graph), but it *could* be tweaked for 3.3 if takkaria thinks the problem is serious. There are constants in make_object that could be used to reduce the escalation via good/great/artifact.

Wand/staff of slow monster still too good. Slowed uniques twice on second attempt, 6 on first attempt (OTH never succeded a second time) (playing H-Troll Warrior). And then usually tough uniques were push overs. I always had time to kill the monsters I had slowed before they sped up. My suggestion is to make it just a bit harder to fail, but shorten the time a lot, so they are only slowed like 4 turns or so.

Certainly it looks like there is still more tweaking to do in this direction.

I really like the new spikes and doors. I think it might really bring out new plays, at least in the early game. But, stuck doors are too hard to bash down. My H-Troll Warrior doesn't really have a problem, but earlier when I was playing a mage he could never open a stuck door. I wasn't really testing it, and he only encountered 2 doors.

That seems WAD. HT-Wa has no problem, mage can't open stuck doors (but has stone to mud spell). Sounds good. How are people finding the mechanism of locking doors? Anyone using it much? Anyone finding it abusable?

Congrats to artes on the doors and spikes code.

[jens]

Regarding Slow Monster: harder to fail, or harder to succeed? I.e. should the wands be more reliable or less?

Less reliable. But I'm not sure if there might be a break point where they suddenly become to unreliable, so it's easier to balance by reducing the power, rather than reducing the control. A very good item that you can only use very seldom, while technically balanced, is IMO not very good for the game.

[jens]

Not on unix it doesn't. What would you prefer it to be called? Or do you want it to prompt for a charname on saving?

The thing is I have never encountered it before, but to be sure I checked, and now I have it on 3.2 as well. I'd prefer that a random name be set on creation if the player did not choose any. PLAYER works as a temporary fix though.

Ok, this could be a bug - are you playing with ai_smart_learn on? And are you referring to non-STUPID monsters (i.e. not jellies / mushroom patches)? If yes to both, please post a savefile which shows this happening and I'll try to debug it.

Yes, birth_ai_learn (and birth_ai_smart) is on. Don't have a savefile at the moment, but I'll try to save if/when I encounter it again. I don't remember which monster it was. But in any case, either they learn and move on, or they are too stupid too learn: I only want to see that message at the most 1 time per monster.

Are you talking about pseudo-ID inscriptions? I rather hope that Eddie's next generation ID will get rid of pseudo altogether, and all these horrible anomalies will go away.

Yes, pseudoID.

Increasing rarity of DSM is easy, though I'm keen to hear from other people if they are agree that DSMs are too common (now that there are fewer egos).

Yeah, this is just a matter of taste. I like DSM being a bit rare, gets more flavourful then. I only got one DSM ego so far, and that was Craftmanship.

Reducing overall numbers of egos and artifacts is trickier. I'd prefer to wait until 3.4, because rewriting the artifact generation to use the "special artifact" route is going to need a whole lot of rebalancing (thanks to Derakon for the baseline graph), but it *could* be tweaked for 3.3 if takkaria thinks the problem is serious. There are constants in make_object that could be used to reduce the escalation via good/great/artifact.

Don't see that it's all that tricky. People have been talking about too many ego's and artifacts since I started reading thses forums, and as far as I have seen no change has been made to fix it. There are two constants in the top of obj-make.c could possibly tweak those a bit. But I think the bigger problem lies in the formulas in the method apply_magic() further down in the same file. Those could pretty easily be adjusted to reduce the total amount of egos and artifacts. And, you can always choose to make a small change in the right direction if you are afraid of shooting to far.

That seems WAD. HT-Wa has no problem, mage can't open stuck doors (but has stone to mud spell). Sounds good. How are people finding the mechanism of locking doors? Anyone using it much? Anyone finding it abusable?

What does WAD stand for? The problem for my mage was that he found those doors before he had stone to mud. Now he could just ignore that part and move on, but I can see situations where you might get stuck in the dungeon because you can't open a door. Would be a shame after ensuring that dungeons are connected. Maybe make the strengths on the door somewhat dependent on dlvl? That way he'd have a chance of having the spell before getting to doors he can't open.

[Derakon]

WAD = Working As Designed.

Mages should get Phase Door pretty early to deal with doors, but priests don't. Other classes should all be strong enough to break doors down, at least early on. I wouldn't mind seeing door strength vary with dungeon level though.

Messaging for monster mana drain got changed somewhat recently; I reported that I was getting messages when they tried to drain my warrior's mana -- this is new. Previously the monster would just silently waste a turn, I suspect. When ai_learn/ai_cheat is off, what would the desired messaging be?

[EpicMan]

Originally Posted by jens
When starting new player, you can leave the name field blank, and get a character without name. This character can then be saved and get's a save file called PLAYER.

Originally Posted by Magnate
Not on unix it doesn't. What would you prefer it to be called? Or do you want it to prompt for a charname on saving?
The thing is I have never encountered it before, but to be sure I checked, and now I have it on 3.2 as well. I'd prefer that a random name be set on creation if the player did not choose any. PLAYER works as a temporary fix though.

This behavior has been occurring in Windows since forever. It was at least in place when Zangband was forked in the 2.X days.

[Magnate]

Mages should get Phase Door pretty early to deal with doors, but priests don't. Other classes should all be strong enough to break doors down, at least early on. I wouldn't mind seeing door strength vary with dungeon level though.

Sure, that would be ok. But I see no problem with priests (or any weak chars without ?PD) being unable to open the occasional door. That's in keeping with a long tradition of "this door appears to be stuck".

Messaging for monster mana drain got changed somewhat recently; I reported that I was getting messages when they tried to drain my warrior's mana -- this is new. Previously the monster would just silently waste a turn, I suspect. When ai_learn/ai_cheat is off, what would the desired messaging be?

Well, first you reported not getting any messages at all, so I added them. It's simple to add a test for "if monster does not already know you have no mana" to avoid the irritation of multiple messages. Note that the birth_ai_smart option is marked as broken so it may give the wrong behaviour and/or messages. But for normal play without it, the "realises that you have no mana" message should only appear once and only if birth_ai_learn is on. If it's off, it will never realise and may keep casting. (If cheating is on, the monster will never cast mana drain anyway if you have no mana.)

[jens]

Note that the birth_ai_smart option is marked as broken so it may give the wrong behaviour and/or messages.

Ah, I have never understood what that meant, a long time I though it ment that the monsters cheated, or became too good, or that it just did not work, so it took a long while before I tried it. Now I play most games with it, because I hope they will be better, but I haven't really noticed any differences...

So, what is required for us to be able to remove the broken sign?

[Derakon]

One thing would be to fix the chain-healing problem with deep uniques. IIRC truly smart monsters are basically unkillable because the odds of them not healing long enough for you to kill them are so close to nil.

It's been a long time since I played with smart monster spellcasting, though.

[jens]

One thing would be to fix the chain-healing problem with deep uniques. IIRC truly smart monsters are basically unkillable because the odds of them not healing long enough for you to kill them are so close to nil.

It's been a long time since I played with smart monster spellcasting, though.

If that problem has been there for a long time, then it's never really been a problem as I see it, just a design choice. Granted they are a bit tougher to kill, but that just means you take them later. On the other hand I can agree that they might become more interesting if they did alternate in some other, but also 'smart' way :-)

[Derakon]

My memory of trying to kill deep uniques with smart spellcasting on was, as I said, that they were unkillable. You're generally lucky to be dealing more than 600 damage/round; their healing spells easily recovered more than that much and they busted those out as soon as they got below maybe 25%.

Again, it's been a long time since I played with smart spellcasting on, so someone should probably verify this, but unkillable monsters strikes me as a good reason to label an option as broken.

[jens]

My memory of trying to kill deep uniques with smart spellcasting on was, as I said, that they were unkillable. You're generally lucky to be dealing more than 600 damage/round; their healing spells easily recovered more than that much and they busted those out as soon as they got below maybe 25%.

Again, it's been a long time since I played with smart spellcasting on, so someone should probably verify this, but unkillable monsters strikes me as a good reason to label an option as broken.

Well, I'll play with it on, and try to see if that is still around. If I remember correctly there were a few that bothered me, but only those which had a casting freaquency of 1 in 2, but even thous get a longer gap between healings now and then...

[jens]

That seems WAD. HT-Wa has no problem, mage can't open stuck doors (but has stone to mud spell).

Tried this a bit more today with a new mage (STR 13). Even if I use only one spike on a door, I can't open it. Encountering a stuck door now and then that I can't open is OK I guess, but not beeing able to open even the lowest level of stuck is not. This means that the value of jamming doors becomes much less, because in several situations were it would be a good play, you also risk locking yourself into the dungeon (I believe this mechanism is something that will almost only be used in the first trip down, so no, mages do not have Stone to Mud).

Another issue is that if there is a break point were it's possible, but takes many tries, then we get the tedium effect of having to bash a lot of times. I didn't check the code thoroughly, but I did see that there is always a chance to bash down doors. But this chance is 1 in 100. That is pretty useless. OK I might not get locked in, but bashing is a very boring command since you cannot use it efficiently (you cannot repeat it automatically).

My belief is that this has been balanced based on super human stats. Which is inherently wrong. Game mechanics for use in the early game should be balanced based on stats 3-18.

[Magnate]

Well, I'll play with it on, and try to see if that is still around. If I remember correctly there were a few that bothered me, but only those which had a casting freaquency of 1 in 2, but even thous get a longer gap between healings now and then...

The solution to unkillable self-healers (and to fixing the smart_ai problems in general) is monster mana. We already have this in a branch on github, so it's definitely something we could bring into 3.4. It would enable a large number of ai improvements, once monsters have finite mana.