Caster of Magic II Bug Reports!

[jhsidi]

Bug:
Endless loop range check error on the next turn -- can't proceed with the game: https://ufile.io/y3dqrnp7 ... Still occurred if I backed up a few turns and played forward. However, when I chose Housing as next build queue item in Cannae (my northern nomad city) the error didn't occur. I was able to build in the city after that and go on with the game, so maybe it was a Cannae issue.

Issue:
Same game: neutral monsters have now destroyed 3! wizards and it's only 1505. I have both rampaging raiders / monsters on, but regardless, I feel like it should be basically impossible for those raider stacks to destroy a wizard fortress. Rjak is the most recent; looking at his garrison (he's on the southernmost continent in Myrror) it looks like the raiders shouldn't be strong enough to take him. Here's the turn before it happens: https://ufile.io/qiqspmsf

[Seravy]

Bug:
Endless loop range check error on the next turn -- can't proceed with the game: https://ufile.io/y3dqrnp7 ... Still occurred if I backed up a few turns and played forward. However, when I chose Housing as next build queue item in Cannae (my northern nomad city) the error didn't occur. I was able to build in the city after that and go on with the game, so maybe it was a Cannae issue.

The lair on Myrror, X=37, Y=10 has two sprites and 10 war bears in it. That exceeds the 9 unit limit so whenever the AI considers to attack the tile, it causes an error.
I need the save files from earlier turns to find the exact moment the lair gained additional war bears. That shouldn't normally be possible.
This is the second time this bug is reported but the first time the save files for the earlier turns weren't available so I couldn't find how the extra units got added.

Same game: neutral monsters have now destroyed 3! wizards and it's only 1505. I have both rampaging raiders / monsters on, but regardless, I feel like it should be basically impossible for those raider stacks to destroy a wizard fortress. Rjak is the most recent; looking at his garrison (he's on the southernmost continent in Myrror) it looks like the raiders shouldn't be strong enough to take him

I disagree : 2 water elementals are much more powerful at range than the one hero, so the raider stack has ranged advantage and can deal significant damage to the other units before melee starts (especially as the defending army is also slow, mostly 2 and 3 movement units.), afterwards it's mostly nagas which means poison damage but the attacking army has poison immunity on 7 of the 8 units. Finally, Phantom Warriors are weak in normal combat but in automatic combat, their tripled attack power (due to Illusion) makes them a major force while being fragile has no relevance (other units share their armor and hit points with them).
Why the AI has nothing better than swordsmen and nagas in 1405 is the question here, they did finish the fighter's guild but haven't actually built any halberdiers. They have the university but no steam cannons.
They aren't a perfectionist which would drop unit building priority and the difficulty level is very high. There might be an AI city production bug or something.

[Seravy]

Update :

2020-09-03
-Fixed bug : The AI failed to find targets for direct damage spells in combat.
-Fixed 09-02 bug : The Ai fails to target combat summon spells.
-Fixed bug : Unit view substracts greyed icons from total incons displaying less cons overall than necessary
-Fixed bug : It's possible to tribute gold you do not have.
-Fixed bug : Starting settlers consume 1000 population each even though they should not.
-Fixed bug : End Turn does not proceed after selecting the first turn research
-Fixed bug : Combat wall spells return "invalid spell" result even if the spell wasn't invalid.
-Implemented attack scaling for automatic combat : attack strengths below 5 are penalized because they have harder than usual chance to overcome average armor.
-Automatic combat attack formula now multiplies attack with (figures+1)/2 instead of figures. This should help simulate the effect of multiple figures being affected by enemy armor more, by only 1 figure units having the default attack power. This means 2 figure units are 25% weaker, 3 figures units are 33% weaker, 4 figures units are 40% weaker and so on. (this modifier only affects attacks, not resistance based abilities.)
-Fixed bug : In strategic combat, units with "other" type ranged attacks ignored figure count for calculating ranged attack strength.
-Added missing feature : Heroism is cancelled from units that reached level 4 already from experience.
-Added missing feature : Guardian adds +6 max pop to the wizard's home race.
-Fixed bug : Raiders are being generated on continents with no cities belonging to any wizard
-Added realm name to spellbook treasure
-In Automatic combat, teleporting and merging units now count as if they had a movement speed of 8.
-Added logging of gold, mana, SP, RP income at the beginning of the turn.
-Fixed bug : AI power distribution decisions that look at own casting skill are wrong.
-Fixed bug : AI early game power distribution strategy had no effect.
-The AI will now avoid building a settler in their fortress city before turn 60 if they already own 4 cities. This should enable the AI to actually use their mandatory early military buildings to produce units.
-Fixed bug : When the AI wants to produce a unit to offset major enemy presence on the continent, it builds a building instead.
-Fixed bug : The AI calculation for having enough free tiles around the city to produce units is wrong.
-Font antialias pixels now count as transparent for drawing a shadow or outline, this should look more like the original game.
-Added save file compression
-The AI will not cast any of the three water move enabling spell categories if they have Wind Mastery in effect.
-The AI will not cast Flight as part of the "overland buff spell" category if they have Wind Mastery in effect.
-After automatic combat, when distributing damage, a unit that already took damage has a slightly reduced chance to take damage again. However if the losing side is neutrals, the unit taking damage has increased chance to take damage again.
(in other words, neutrals tend towards losing more units to combat damage while non-neutrals tend to lose less than as if the damage was distributed randomly. This should enable using the sprites vs land units tactic with automatic combat.)

So yeah this definitely explains the weak AI, it didn't cast combat spells, it didn't build units and it even assigned its power wrong.
A reminder, I will still need the save files from all previous turns to find what caused the 10 war bears in that lair.

[jhsidi]

A reminder, I will still need the save files from all previous turns to find what caused the 10 war bears in that lair.

Sorry, that game was having issues so I started a new one. I'll keep the autosaves in the future when running into a bug of that sort.

[jhsidi]

I disagree : 2 water elementals are much more powerful at range than the one hero, so the raider stack has ranged advantage and can deal significant damage to the other units before melee starts (especially as the defending army is also slow, mostly 2 and 3 movement units.), afterwards it's mostly nagas which means poison damage but the attacking army has poison immunity on 7 of the 8 units. Finally, Phantom Warriors are weak in normal combat but in automatic combat, their tripled attack power (due to Illusion) makes them a major force while being fragile has no relevance (other units share their armor and hit points with them).

Yep, I agree with your analysis of the battle. To clarify I meant that it shouldn't be possible for neutral raiders to take an enemy wizard's fortress no matter what the strength imbalance is, because it's unfun for the player to lose their opponents in that way. Since you're figuring out a production bug that seems like most of the necessary solution, but maybe the enemy fortress could also get a strength buff against neutral stacks?

[Seravy]

Yep, I agree with your analysis of the battle. To clarify I meant that it shouldn't be possible for neutral raiders to take an enemy wizard's fortress no matter what the strength imbalance is, because it's unfun for the player to lose their opponents in that way. Since you're figuring out a production bug that seems like most of the necessary solution, but maybe the enemy fortress could also get a strength buff against neutral stacks?

Probably there should be something to simulate the effect of fortress lightning in fortress battles instead, maybe having a fortress could raise the "direct damage" effect level of spent combat MP by a high amount.
But unfortunately, with more accurate automatic combat comes the increased possibility the AI loses the battle because they are weak against the stuff attacking (like poison immunity vs ghouls or nagas, or possibly missile immunity against slingers/bowmen etc, although missile immunity is rare on monsters so it's less of a problem. A stack of sprites against a gnoll wizard with no ranged units though could also be very bad. But in general if the doubled monster options are not turned on, the stacks shouldn't really be powerful enough to be a problem even if the AI gets unlucky with the type of units attacking them.)

[jhsidi]

Endless loop bug when a war bear stack attacks my magic spirit: "Access violation at address 0065AA8A in module 'Caster.exe'. Write of address 000002A5." 

Here's the save file: https://ufile.io/v4ogxqgi

edit: actually it happens anytime anything attacks one of my stacks or cities, regardless of whether it's a wizard or a neutral

[Seravy]

Endless loop bug when a war bear stack attacks my magic spirit: "Access violation at address 0065AA8A in module 'Caster.exe'. Write of address 000002A5." 

Here's the save file: https://ufile.io/v4ogxqgi

edit: actually it happens anytime anything attacks one of my stacks or cities, regardless of whether it's a wizard or a neutral

It's crashing on your wizard's name. It's too long and tries to draw the text outside the screen. I guess we have to reduce the maximal name length further.

[Seravy]

Updated :

2020-09-04
-Implemented character limit to text input.
Wizard name : 10 characters
Artifact name : 18 characters
City name : 14 characters
Hero name : 10 characters
Save slot name : 20 characters
-The AI now prefers targeting the lowest speed unit with Wind Walking (as it now sets a minimal movement speed)
-Automatic combat at the fortress city raises the "Direct Damage" spell level of the defending wizard by 10.
-Combat nature nodes are now animated for more visibility
-Fixed bug : when saving, the uncompressed temporal file is not deleted
-The Info menu can now be cancelled by clicking outside or pressing the Esc key.
-Fixed bug : Range check error when Unicorns appears for research due to description too long.

[jhsidi]

I haven't been able to start a new game with this latest version. The game stalls out after selecting a flag color. Before, I think it took around 15 seconds to start; now, I've waited up to 5 mins but it stays stuck on the flag screen.

I've tried selecting different options at the start, including the game and continent sizes, but none of them make a difference.

edit: So I tried deleting the folder and unzipping the file again. Now map generation works fine. I misremembered the 15 second thing -- it's actually nearly instantaneous. I did do one odd thing with the game the first time I unzipped it: I ran it once, then stopped the game (before generating a map) and moved the files into a new folder and restarted it. So, perhaps the act of moving the game files was the culprit.

Anyway, since I'm thinking about it I'll play with map generation a bit.