OK, that endturn issue doesn't happen anymore. Here's a new one instead: https://ufile.io/5m1faie1
|
Caster of Magic II Bug Reports!
|
|
September 21st, 2020, 10:35
(This post was last modified: September 21st, 2020, 12:16 by Seravy.)
 (September 21st, 2020, 10:27)jhsidi Wrote: OK, that endturn issue doesn't happen anymore. Here's a new one instead: https://ufile.io/5m1faie1 Even before the first AI had a chance to move anything, this appears in the log : "Tile has more than 9 units! 1,27,13 Tile has more than 9 units! 1,12,57 Tile has more than 9 units! 1,12,57 Tile has more than 9 units! 2,38,37" So...could you please find the last turn before the current one where that wasn't the case so we can see what move resulted in that? Meanwhile I also started a test game (finally my todo list only had missing art and graphic improvements left which don't affect playtesting) and found one reason for the AI cities having no defenders : Housing. As simple as that, the AI is set to build housing if the terrain/etc makes it beneficial which can be anywhere from 0 to 7 population. This leaves the city undefended for those turns which isn't that many turns actually, in 1402 the first AI city finished doing it and had 2 buildings which means probably about 10-12 turns spent on housing at best. However, in exchange the AI can start to work with a city that has that much higher population so it should fill with units faster afterwards. In theory. Now, that's the theory, but the reality is, the housing ignores the AI production bonus which makes it less efficient (housing discards production). It also ignores the "Buy sawmill first" strategy but that's fine because housing is more efficient until the calculated population... but only in theory, in practice that would only be true if the housing output was divided by the AI bonus ratio. Of course the AI population growth ratio increases it but that ratio is lower. So the AI discards 130% production for 110% population growth for example and that 20% difference is ignored in calculating efficiency, in the end the AI does housing as long as a human player should. But there is another problem. If difficulty is not that high, the AI will be short on gold and fail to buy their sawmills. Why? Because Housing is set up to allow buying the marketplace and magic market first from gold if able, as those two can produce resources during housing and ideally pay back most of their costs by the time housing is done. Except the problem is the AI will keep buying these in the next coming outpost whenever it has gold so it never has any to spare (and magic market pays the investment back in power) for buying the sawmill in their first one. Not having the sawmill until it is built through production delays those towns by another 5-6 turns. This probably doesn't happen at Expert, maybe Master or higher thanks to the gold income bonus and higher starting gold but definitely seems to be a thing on Advanced. Overall, due to housing (and purchases related to it) the AI might be stuck with just the first unit (summons can't really happen on the first 2-3 towns because Summoning Circle research isn't free and while units might come from the fortress, only if the fortress is overflowing with units already and that too isn't happening on lower difficulty before turn 15-20.) ...so the AI might be stuck for an additional 10-15 turns with just a swordsmen in their cities, which isn't all that much but definitely enough to lose the city and it'll look much worse than it actually is because the city will be pop 4-5 and have a marketplace and magic market due to the housing. Also, Housing vs building the sawmill is better, but compared to outright buying it, not necessarily. However I do remember the housing did boost AI efficiency by noticeable levels so I can't say that it's an outright bad strategy, it just need more fine-tuning. In other words, the early game optimizations for booting up economy in new cities prevent those cities from building the initial 4 swordsmen for quite a while if, and only if, the city was built on quality terrain that allows relevant amounts of housing. I'm not sure if using Housing is really worth the extra risks for the AI although 4 swordsmen don't make much difference either and better units won't be made in the city for a while and housing actually helps with that. Maybe require Expert difficulty for AI housing? That should at least solve the gold issues with buying the sawmill. Or even better, require Expert difficulty for buying markets, and leave Housing for all difficulties? Low difficulty can't afford buying sawmills either way so letting housing run and hold onto the gold until it's complete then buy the sawmill might be best. Marketplace gets the gold back though so maybe only limit the magic markets? ..I am kinda wondering why that wizard had no monsters to pull into cities though. The same Life/Chaos in this next test run summoned enough Hell Hounds to fill their fortress by turn 12. ...oh it did have them. It was only enough to send some to the first city, the rest haven't got any by the time I ended playing. The wizard with all Life books had no units though. Fortress builds fighter's guild or settlers, new cities are busy housing and getting the base economy up, and the wizard cannot summon. That would be a recipe for a disaster except the buffed single swordsmen in the cities probably pretty decent. Yeah... swordsmen with 9 shields, 2 hp, and 7 swords, that's kinda better than 2-3 hell hounds, I guess Life wizards are fine.
I made a mess of all the save files and it looks too hard to dig through now. But I'm sure I'll hit this bug again soon, it seems to happen frequently.
Question -- how do I navigate through the log easily? I don't see anywhere in the log that the exact turn number is named and I'm not sure how to know what turn I'm looking at when I'm scrolling through it. Actually, the turn number isn't even stated in the game, it's just month and year. For bug testing it would be nice if the turn # were added alongside the FPS. Unless I'm missing it somewhere?
This was supposed to be tomorrow's update but might as well post it now :
Quote:2020-09-22 (actually 21 still) (September 21st, 2020, 10:35)Seravy Wrote: In other words, the early game optimizations for booting up economy in new cities prevent those cities from building the initial 4 swordsmen for quite a while if, and only if, the city was built on quality terrain that allows relevant amounts of housing. Really my only problem with small garrisons in outposts and growing towns is the rampaging monsters. The AI can't defend against the player with this kind of outpost unless there's a big change to ensure full strength stacks go with every outpost. But my observation has been that a huge number of isolated cities get wiped even by weak stacks of monsters, and that has to be having a negative effect on the AI's ability to snowball. It's so frequent that there'll be places where I see 3 or 4 ruins clustered because the AI just keeps sending settlers to the same spot the last city got burned down. Would it be OK to adjust the likelihood of spawned monster stacks so that they're more likely to target larger cities? This would benefit the player as well, obviously. But that's OK on normal difficulty. And on high difficulties, it wouldn't be that big a benefit -- the AI can always fully garrison a decently sized city, but as the player I'm resource starved and usually shoestringing the garrison as much as possible, so the rampaging stacks are a threat to pretty much any town I own. There's good lore behind it, too; the more impact a settlement makes on the natural world, the more likely it is to draw attention.
You may recall I reported that enemy hero items don't drop in battle after they're killed. You asked me to find an instance of it, but I actually haven't seen a single hero with an item since then, but here's something you can test: you don't get back your own hero items if they die normal deaths in battle.
You can test it out in this battle, for me Greyfairer died to some swordsmen but I didn't get his cloak back. Sorry, the battle is a pain in the ass, Ariel is Life / Guardian. It's the city adjacent to the only active stack: https://ufile.io/o7bfgpey Btw, could Discipline, Mislead etc just get a placeholder blank icon for now? I had it written up as a bug that all the units had an extra move until I remembered the new spell.
September 21st, 2020, 23:05
(This post was last modified: September 21st, 2020, 23:06 by Seravy.)
Quote:You can test it out in this battle, for me Greyfairer died to some swordsmen but I didn't get his cloak back. Sorry, the battle is a pain in the ass, Ariel is Life / Guardian. No problem, we have Shift-G now. Oh, wait, I better change it to activate during combat otherwise the hero can't die. Quote:Btw, could Discipline, Mislead etc just get a placeholder blank icon for now? I had it written up as a bug that all the units had an extra move until I remembered the new spell.I've added an icon for the unit enchantments in one of the updates last week. Are they still not working? |
