Caster of Magic II Bug Reports!

[jhsidi]

- On "All units retreat exhausted" results I'm losing 1 fame sometimes. Usually it's when I kill all of their units but one invisible one, and I'm not taking any losses. Maybe the opponent is supposed to be losing the fame?

Did you summon anything? Combat summons do count as lost units unless we change the rule, so if you aren't winning the battle,and you summoned at least 12 experience worth of creatures, you lose 1 fame.

Ah, that would be it. I'm summoning phantom beasts to deal with night stalkers.

Yeah, that rule makes no sense to me. They're disposable units by definition. Just seems like a punishment for summoning-heavy players.

[jhsidi]

Wanted to give you some more examples of cities being under-garrisoned on high difficulty levels, since it seems like an issue in all the games I've played. It's from this file: https://ufile.io/xdbhsva2

Arcanus:
17x12: Sharee's city, 1 unit (not counting magic spirit), she has owned it for maybe 8 turns
11x16 and 11x12: Raven's cities, 4 units each, he has owned one for 20+ turns and the other probably a bit less
32x29: Kali's city, 2 units (not counting magic spirit), she has owned it for 10+ turns
30x25: Freya's city, 3 units, not sure how long owned
34x24: Ariel's city, 2 units, not sure how long owned

Myrror:
27x27: Horus's city, 0 units!, not sure how long
24x32: Horus's city, had 3 units before I conquered it
31x24: Horus's city, 1 unit, not sure how long
(I don't want to look at the rest of the plane since I just broke in).

Horus has a doom stack right by his empty cities. I understand that the game logic forbids breaking up or reassigning these stacks but there's still something wrong here. Another observation: he has 12k gold and 3.5k. Some of the Arcanus wizards have similarly strong economies. Yet the cities are empty.

This is with Monsters Gone Wild turned on, which is part of the problem. It's got a 1.3x score modifier but honestly, it makes the game much easier because the AI can't reliably expand with the monster stacks spawning.

Ideas:

• - Give a higher priority to strongly garrisoning cities, even new ones, versus building doom stacks. As long as wizards are consistently losing cities, they're also burning resources on building settlers and buying basic structures in freshly built cities. This cycle can go on for years. Garrisoning the cities would take some time but in the medium to long term, it would result in more resources available for these wizards.
  
• - While the wizard has under-garrisoned new cities, put a halt on the expansion AI. As in: the wizard should NOT continue building settlers and sending them out. Instead, they should use that resource for building units and routing them to the settled city.
  
• - Reduce the likelihood of wizards attempting to colonize other continents. I see wizards sending out settlers to further locations even for their first 2-3 cities. It works out very poorly for them because they can't direct units there. They should prefer a poor 10 population slot adjacent to their capitol over a 18 population slot on a continent that's 5 ocean tiles away.
  
• - If the wizard has money for it, buy military buildings and then units so long as funding is available.
  
• - Consider changes to the stack building mechanic. I don't know the details well enough to suggest anything specific here, but there is definitely a problem. Early game, for example, I see stack after stack of 6-9 nagas roaming all over the world while the wizard leaves his cities defended by a weak city units. Jafar did that this game, he had about 8 doom stacks of nagas while even his fortress had a couple units of swordsmen in it.
  
• - Maybe consider putting more wizards on Myrror (assuming the player is on Arcanus)? I'm not sure about this, but the 3 wizards are not doing well in my games. I think they're spending all their time desperately trying to expand and failing at it because of aforesaid failure to garrison. Maybe with one or even two more wizards there, they'd consolidate more and one would have a better chance of becoming an empire.
  

[Seravy]

I'll look at that save file but have you actually looked at the contents of the log file to check what the AI was building in the cities? It's possible they choose the wrong things but also possible the production income is miscalculated so the units never finish building.

Klackon city : this place has 6 rebels and 4 farmers.  It has no fighter's guild yet. My guesstimate is they are building a fighter's guild.
There should be at least 2 more swordsmen though - the mandatory "minimal 4 units" is before the buildings I think.

The two grey cities : These should be pull units until they have 6 but there isn't really anything on the continent the AI could pull, they only have 3 units on the whole continent. The units inside the cities look like things that were pulled previously, 
...actually, I clicked on an empty tile next to the city and it brought up a unit view window with zero units. That's a bug but it indicates an invisible unit on the tile. Those cities might very well have additional night stalkers inside.
So I added a debug key to show invisible units and the cities did have stalkers so they are fine.

32x29 : this too has stalkers


30x25 : definitely needs more units. The AI doesn't have anything to pull inside though anywhere near, except a single naga (that should probably enter the city next turn, or not I don't remember if intercontinental units are pulled since the previous updates or not)
The city did manage to produce 1 spearmen already, maybe it's still working on the second.

34x24 : looks freshly conquered, this wizard is at war with the closest neighbor

27x27 : This has no smithy. It must build one before it could produce a unit.
However, I believe this is completely empty because the doomstack conquered it 2 turns ago and left it this turn. In theory the AI should summon immediately the same turn to fill it, but if they were busy casting another, longer spell, they couldn't do that and missed the chance. Compared to CoM I where the AI would have simply razed the city, this is still better.

31x24 : Same here, without Smithy, units cannot be built.
The absence of smithy implies both of these were freshly conquered, probably by that doomstack.

- Give a higher priority to strongly garrisoning cities, even new ones, versus building doom stacks.

For "normal" stacks, the priority is pretty much the highest possible - 8 units must remain in the city, only one is allowed to leave. For what I see in the code I think this rule applies to doomstack material units as well, unlike in CoM I though you can probably find much more detail in the posts where this was discussed (somewhere in the older brainstorming thread I guess?)

However, the intercontinental doomstack "rolling out" of the city after it was built is an exception. That happens after it took a city and continues to the next target as well as if the city got filled up to 9 intercontinental units by the AI summoning to the frontier city.

So the only way we can affect this is changing the ratio between summoning to garrisons or frontier cities, if that's what you are suggesting, please find the related discussion about AI frontier city and doomstack handling and we can continue that discussion.

- While the wizard has under-garrisoned new cities, put a halt on the expansion AI. As in: the wizard should NOT continue building settlers and sending them out. Instead, they should use that resource for building units and routing them to the settled city.

I don't think the AI can build a settler before the mandatory "have 4 units" in a city. If you mean a new rule, you must actually define what it is (under -garrisoned is subjective. The AI needs exact numbers - how many or what value units, etc)

- Reduce the likelihood of wizards attempting to colonize other continents. I see wizards sending out settlers to further locations even for their first 2-3 cities. It works out very poorly for them because they can't direct units there. They should prefer a poor 10 population slot adjacent to their capitol over a 18 population slot on a continent that's 5 ocean tiles away.

Unless the AI actually plays Lizardmen or Draconians, this rule would simply waste the water walking/wraith form enchantment on the settler - and if they do, sending troops over should be less of an issue.
However, adding a rule that says settlers have a bonus priority to their own continent if they cannot cross the ocean without a ship would make sense. Question is how much bonus priority? 4 max pop worth? 6? 8?

- If the wizard has money for it, buy military buildings and then units so long as funding is available.

Production selection and decision for spending money are separate.
We can make the AI prioritize producing military buildings and units higher than other things - when? Simply because they have money? Why? Buying economy buildings is much better if they aren't fighting a war. So I think changing this would be a dead end, the AI should pick what it builds depending on what it needs, not depending on money.

...or we can make it spend gold more willingly on those things instead of others. But simply spending gold on units all the time is still bad strategy. Gold should be spent on what is urgent to complete. So we need to somehow define that.

Currently, units have double priority to buy below 6 garrison size. Quadrupled below 4 size and 8x priority to any production below 2 units. It also has increased priority if special ores are present, or enemies are on the continent to all production. What else should I add?

- Consider changes to the stack building mechanic. I don't know the details well enough to suggest anything specific here, but there is definitely a problem. Early game, for example, I see stack after stack of 6-9 nagas roaming all over the world while the wizard leaves his cities defended by a weak city units. Jafar did that this game, he had about 8 doom stacks of nagas while even his fortress had a couple units of swordsmen in it.

It was explained in details in the mentioned thread, find it and revive/continue that discussion if you disagree with the current solution.

A finishing thought - if AI performance is so weak across the board, there probably are more critical bugs causing it. Pay attention to logs - maybe the cities sometimes don't produce the correct amount of hammers, or the AI fails to spend gold?

[jhsidi]

A finishing thought - if AI performance is so weak across the board, there probably are more critical bugs causing it. Pay attention to logs - maybe the cities sometimes don't produce the correct amount of hammers, or the AI fails to spend gold?

Could this be a bug? The city owned by Horus at 31x24 has remained static for several turns, with 1 storm giant as garrison and no new cities or units being built. Looking at the log for the city, Pyre, it's producing something but apparently he's waiting for it to build rather than rushing it. Here's the log file: https://ufile.io/vxkl6yzc

I'll look for that older thread.

[Seravy]

Well one thing I do see from that log is, the city produces 1 hammer.
Which explains why nothing gets built.
It's not a bug though - 2 rebels, 1 farmers, no smithy and no sawmill does get this result.
The AI however has 12k gold and it's unthinkable to not buy a smithy with that much money so this might be a bug in AI buying production. (In fact, over 10k gold the AI should be buying everything instantly.)

[Seravy]

Update :

2020-09-16
-Fixed bug : Peace treaty counter was set to the wrong amount.
-Fixed bug : When invisible units move in combat, their position is incorrectly set
-Added missing feature : AI Dispel Magic target selection function
-Fixed bug : Dispelling Confusion doesn't clear the "confusion effect" variable
-Fixed bug : Dispelling Web dooes not clear the remaning web strength variable
-Fixed bug : AI can do diplomacy with other AI if at most one of them is banished.
-Fixed bug : Elemental Armor is missing from the "Spell ID to enchantment ID" conversion table
-Added Ruler of Heaven global enchantment image (by Slingers)
-Global enchantment and mirror frame drawing priority now matches original game.
-Added Call the Wild global enchantment image (by Slingers)
-Added unit icon for Misfortune
-Added unit icon for Destiny
-Added unit icon for Healing Aura
-Fixed bug : Right clicking an invisible enemy stack shows an empty unit view window
-Added logging for AI buying decisions
-Fixed bug : The AI can only buy production in cities where completion takes less than 1 turn instead of more.
-Added debug key : Shift I to show invisible units. (note that this does remove the invisibility ability from the unit, including all of its effects.)

That gold spending bug might explain most of the garrison issues - if the smithy, sawmill, etc aren't bought early on, cities can't produce a thing unless on good enough terrain for housing.

Some explanation on how to interpret the logged buying data.
The AI buys if remaining cost*priority is less than the AI's gold available and the remaining turns would otherwise be more than one. So lower priority means more likelihood of buying it.
Random chance is not involved, and priorities are not compared with each other - the AI will simply look at their cities from oldest to newest and keep buying where it can, skipping those cities where it is false because gold is already lower than priority*cost.

The default buying priority of everything is 100, except for
"BSawmill,BfighterGuild,BAlchemistguild,BWizardsGuild,BAmplifyingTower,BLinkingTower" are 4
"BMarketplace,BMagicMarket,BMonument" are 1 if the city is below pop 7
and everything is priority 1 if the AI has over 10k gold.

( Edit : This latter condition could use some refining maybe? Hard to say because pouring gold into the already developed cities to pump out more units, and pouring gold into smaller cities to get the economy buildings up are both good. Maybe this could use a condition depending on the AI's diplomacy status? Maybe it could look specifically at the human player only? (as wasting economy potential on the AI's infighting is bad, it's better if the strong AI simply cannibalizes the weaker one.) A treaty with the human player isn't very reliable though, they can break it or decide not to get one despite maintaining peace. So this would need the AI to somehow predict the player's plans. )

Then modified by garrison size, number of own vs enemy cities on the continent, Availability of Adamantium, Mithril, Orihalcon, and Chaos Rift.

[jhsidi]

Bugs:
- Enemy hero equipment is not given to the victor, no matter how the hero was killed.
- AI wizards keep casting mislead on additional targets even if it was successful already
- Counter Magic still doesn't seem to do anything; Philosopher's Stone does work though.
- The health bar within the unit stats panel at the upper-right of the screen during battle is usually incorrect for high hp single figure units. E.g. if the unit is at 1/4 health, the health bar has 0% fill.
- Cracks Call can be cast on water. Also, not sure if it's related to how regeneration works, but when it hit my great lizard, it did not regenerate the lost health after battle.
- During battles that take place on a tower tile, the tower graphic is not there.
- I haven't yet seen any banished wizard come back, even many turns later; the spell of return may not be working at all.
- When neutrals banish a wizard, the message still says "Raiders destroy Name" instead of banish name.
- Allies who want you to join a war against their enemy will ask every single turn, and relations take a huge hit every time you say no; I've gone from peaceful to hate with an ally in under 10 turns.

Notes:
- Not sure if it's a bug, but spell save equipment doesn't seem to affect gate of hades. e.g. -4 spell save wand + gate of hades on units with 11 resistance results in them never taking HP damage. IMO it's probably better if spell save equipment doesn't affect area spells as it'd be overpowered, just reporting it in case it is supposed to have an affect.
- Also, gate of hades animates on units even when it has no effect on them (due to high resistance). A bit slow to watch and not consistent with how it used to work.
- It would be nice if razing huge cities were faster. Seems like it reduces 3 population each turn? That's a long wait for a metropolis-sized city. Also, doesn't make much sense practically; my fearsome army really takes 6 months to kill off a local population?

[Seravy]

- Not sure if it's a bug, but spell save equipment doesn't seem to affect gate of hades. e.g. -4 spell save wand + gate of hades on units with 11 resistance results in them never taking HP damage. IMO it's probably better if spell save equipment doesn't affect area spells as it'd be overpowered, just reporting it in case it is supposed to have an affect.

Not a bug, Spell Save only works on the spell's effect at the time it is cast, it doesn't affect any triggered ability on it if it's an enchantment.

- Also, gate of hades animates on units even when it has no effect on them (due to high resistance). A bit slow to watch and not consistent with how it used to work.

We can consider this a bug, it does make no sense to animate if it's zero damage.

- It would be nice if razing huge cities were faster. Seems like it reduces 3 population each turn? That's a long wait for a metropolis-sized city. Also, doesn't make much sense practically; my fearsome army really takes 6 months to kill off a local population?

Razing is intentionally slow, that's why it's no longer "end of combat" instant razing. If you want to get rid of the town, you have to be able to defend it long enough to give the enemy a chance to counterattack. 3 a turn removes a maximal size city in 8-9 turns which is completely reasonable. It takes like 150 turns to build one...

You aren't killing civilians, if you did that, people would revolt everywhere once they hear about that. You just...make it inconvenient to live in the city because, well, buildings get sold, soldiers no longer protect them from monsters, maybe they aren't given food... or maybe you're selling them as slaves, you are getting money for razing the city afterall. Better leave this to the imagination of the player.

[jhsidi]

Re: razing, I dunno. Your logic makes sense but it's just really boring to wait for 8-9 turns for a city to go away. By 2-3 turns after I've taken the city, I've handled any threatening armies nearby; by 5-6 turns the battle is long forgotten. Keeping in mind that at the stage of the game you're razing a 25 pop city, turns take a while and 8 or 9 turns just feels like eternity. Unlike early game where it goes by in a couple minutes.

Bugs:
- Several strange events conspired to destroy Tauron; his fortress Freiburg is at 5x22 on Myrror. First, he fought a neutral monster stack at the end of this turn: https://ufile.io/kygrao84 ... The stack somehow destroys his entire garrison, although it doesn't make much sense that it would be able to do so. It also completely wrecks the city. Two turns later, a neutral stack of draconians appears adjacent to a Chaos node and, over the following turn, navigates to his city. Over this time, he doesn't manage to build or summon a single thing, leaving his capitol completely empty. Here's the turn before they destroy him: https://ufile.io/2m5r7e9s ... it's unclear to me why neutral racial raiders have started appearing despite there not being any neutral cities on Myrror; although it seems to have started around when I cast Call of the Wild. Also, a special provision probably needs to be made to navigate any available units in range into the fortress if it's <X units, this is simply too dangerous a scenario for wizards. Also, here's the log file from the turn after Tauron is banished: https://ufile.io/dwzkho5j ... maybe I don't understand what's going on here, but it looks like he's buying buildings in Freiburg like sawmill and city walls, yet I don't see those in the city the turn before he's destroyed. Also, he summoned various creatures; could there have been a special provision to summon those creatures to his fortress? That might have saved him.
- Also note that even with the toned down monster stacks, wizards are getting wiped out left and right by monsters at this point in the game. Here's Oberic dying to a stack that's actually pretty nasty, capitol on Arcanus at 42x47, the stack includes gorgons which is probably why it won. Incidentally, both Oberic and Tauron lost because their fortresses are full of only great lizards, so maybe those shouldn't be quite as high priority; or, the wizard fortresses should just get an autoresolve buff for lightning like you mentioned before: https://ufile.io/vfghwxm9
- I keep getting offers for blank items like "Trinket of Mediocrity" for 20 gold. This happens most often when I have low gold, but has also happened when I have 1500+ gold. I've got almost 200 fame so offers are coming quite often, but a lot are for these Mediocrity items.
- I took a dark elf swordsman and water elemental into a battle against a magic spirit, but only the swordsman appeared. After the battle, the water elemental was on the map tile. I couldn't get this to happen again on reload, so it's just something to look out for, or perhaps you'll have an idea of how it could happen.
- I attempted to use animate dead on a unit of unicorns, but the window that appeared was empty and prevented anything else from being clicked. You can try it with this save game, take Ravashack with the two great lizards into the lair at 51x14: https://ufile.io/2nq2j0gg
- Even with an overland casting skill of >240, I can't raise supernatural units as undead. You can try it on the above file, Ravashack is adjacent to a tower with djinn, you can use vertigo / web to tie one down and kill it with an iron skin enhanced zombie. Even if the zombie makes the kill alone, the djinn won't raise.

BTW, I was apparently wrong about counter magic not working. I used it in a battle and it worked fine. Very strange, I used it in something like 3 battles in a row and it let through every spell, even the low cost ones. RNG weirdness perhaps?

[Seravy]

- Enemy hero equipment is not given to the victor, no matter how the hero was killed.

When you see an enemy hero with an item to take, please post the save file. Otherwise this'll wait until I play a test game of my own and encounter one.

- The health bar within the unit stats panel at the upper-right of the screen during battle is usually incorrect for high hp single figure units. E.g. if the unit is at 1/4 health, the health bar has 0% fill.

Nothing seems to be wrong on the hp bar but the unit's health display is pretty much unreadable, the taken hit points in the back row are almost the same color as the remaining hit points. I'll make them a darker grey.

- Cracks Call can be cast on water.

Wasn't able to reproduce. All sea tiles are marked invalid and if I click on them anyway, I get the red windows that says it's not valid.

- I haven't yet seen any banished wizard come back, even many turns later; the spell of return may not be working at all.

Casting progress should be in the log file. Does the casting not progress for the wizard, or does it progress but when complete the spell fails to work?

- Allies who want you to join a war against their enemy will ask every single turn, and relations take a huge hit every time you say no; I've gone from peaceful to hate with an ally in under 10 turns.

While this was a bug, the chance in the code is also not set to the same amount as CoM I.

CoM I had a 1/40 chance of demanding a war declaration which is quite low. However the CoM II code has 1/2 chance which is way too high, probably a placeholder number.
Not sure what amount we want to use. As there are more players making it easier to get allies, and one of our main goals is to guarantee enough players are eliminated before the endgame spells come into play, the system definitely should be stricter about this than the previous game.
The original 1/40 chance was roughly an equilibrium - if you denied once every 40 turns then your relation level roughly stayed at the same amount - but in an unpredictable way - get the demand 2-3 times in a row and you have no choice but to accept it or lose your alliance, which forces players to offer tributes after the first depend to repair the relation damage done.

If we make it more frequent, tributes will still solve the issue in theory but in practice it will not work forever and sooner or later there won't enough enough spells or gold to tribute. However, it is reasonable that if you aren't helping your ally with military then you must aid them economically. So I think that's a change in the correct direction.
Maybe a 1/10 chance to demand per turn?