Traits that are bugged in 1.2 with fixes. Part 1: Scarred, Farmer, Trader

[Pode]

^
again, i think the julii temple (ceres) is a fertility temple, giving fertility bonuses. it is not a farming temple.

Tai, think about the word cereal for a minute

[Pode]

Correct me if I'm wrong, it happens a lot, but when a dip opens talks with your general, if you counteroffer and they reject that outright, doesn't the full diplomacy menu, including bribe, come up if you try to make a new offer? Is that the situation that smoothtalker was meant for? Defensive bribing of a diplomat that has approached your general?

[tai4ji2x]

Tai, think about the word cereal for a minute

yes, but in the game it acts as a temple of fertility. look inside EDB.txt

only dacia, carthage, and macedon have farming temples.

[Pode]

Well, at least I did just say that I was wrong a lot, now we have proof. Mea Culpa

[Bob the Insane]

Correct me if I'm wrong, it happens a lot, but when a dip opens talks with your general, if you counteroffer and they reject that outright, doesn't the full diplomacy menu, including bribe, come up if you try to make a new offer? Is that the situation that smoothtalker was meant for? Defensive bribing of a diplomat that has approached your general?

Good point, so the trait is having a positive effect as it is, cool...

Do you think it would be reasonable to mod this trait to make it applicable to diplomats as well as family members? Or could make the diplomats overpowered? Considering how much more expensive bribing is in 1.2 I don't think applying this to diplomats would be unbalancing...

[Sinner]

As I recall family members can counter-bribe when a diplomat contacts them, they just can't initiate a bribe themselves. So Smoothtalker would help in these circumstances.

[therother]

Counter-bribe? How does that work? I've both bribed family members and had family members bribed, and I've never either lost my diplomat to the other side or received the option to bribe the opposing diplomat, although I admit I haven't had too many family members bribed. Perhaps they didn't have the necessary skill?

[Bob the Insane]

As I recall family members can counter-bribe when a diplomat contacts them, they just can't initiate a bribe themselves. So Smoothtalker would help in these circumstances.

This should be quite easy t test for (at least whether the general actually picks up the trait is asy to test)...

With the modified trait file I am using if I can get myself into the position of a diplomat contacting one of my generals to negotiate something, if I counter by bribing the dilpomat and succeed the general should automatically get the trait (if it is all working)...

[Sinner]

theother, when a diplomat contacts you, eg trying to negotiate a trade agreement, you have the option of sending a counter-proposal. Bribery is one of the options on the list, so instead of negotiating the deal the AI wants you 'steal' his diplomat.

[Olmsted]

I would like to see this thread get a bit back on topic. Perhaps it was a mistake, but I read all 5 pages of this thread. Now, having read all that I am confused as can be. What changes should I make to fix the bugged traits? I would really appreciate a final version of the changes that takes into account what has been said for the past five pages. Thank you for your time.

[Bob the Insane]

I would like to see this thread get a bit back on topic. Perhaps it was a mistake, but I read all 5 pages of this thread. Now, having read all that I am confused as can be. What changes should I make to fix the bugged traits? I would really appreciate a final version of the changes that takes into account what has been said for the past five pages. Thank you for your time.

I think the reason for your confusion is because it has not been decided or agreed what the final solution should be...

We seem to have broken down into a couple of basic groups. The first is to simply adjust the threshold values for the traits in question to balance out the problems. The second is to alter how the triggers that generate these traits work.

The first is justified in that it is the least intrusive makes only a minimal change to improve the way the game plays. The second is justified if it is found that the triggers that power the traits are fundimentally broken and no amount of balancing the threshold values will fix things...

In some instances this is obvious like for the Scarface issue where the trigger is fundimentally flawed and needs to be fixed.

In other cases like GoodFarmer and BadFarmer it is less clear as we have not fully agreed how common these traits should be. The only thing that is obvious is that the poor manner in which the BadFarmer trait is triggered combined with it's bizaarly low threshold values makes the BadFarmer trait rediculously common...

I for one am a fan of altering the main trigger that generates the BadFarmer trait....

[Hobart]

Can someone clarify for me what GeneralHPLostRatioinBattle actually measures? Is it (hitpoints lost)/(total hitpoints) as the name sugegsts?

I only ask because I thought from some other posts that it might in fact be (hitpoints left)/(total hitpoints). It would seem from Gregba's suggested fix (some pages back) that he believes it is this latter formula. I say this because he suggests you should get scarred if GeneralHPLostRatioinBattle is < 0.3. iirc that is, I haven't checked ...

This (obviously) affects what the scarred trigger will do. I gather from discussions that the consensus is that you should get scarred if you lose 70% or more of your hitpoints. Am I right about that?

Anyway, can someone set me straight on what the this function actually returns, and I giess I can make my mind up on where the break should be. I, too, am sick of getting horribly scarred people who lose influence as a result, when the most they have done is ride down a few routers.

[player1]

And I thought that concensus was
GeneralHPLostRatioinBattle is < 0.7
or scarred after losing 30% of HP.

And that's probably what CA wanted, but reverted the function from HPlost to "HPleft".

Giving it to just those with less then 30% hp would make it very, very rare trait.

[Olmsted]

Thank you. I will make that adjustment to the battle scarred trait in my file.

[tai4ji2x]

i personally think only losing 30% is still too little (ie, still too easy to get the trait). i set the trigger to 0.6 (%40 HP lost).

[player1]

i personally think only losing 30% is still too little (ie, still too easy to get the trait). i set the trigger to 0.6 (%40 HP lost).

Well, there are multiple level of that same trait, so it takes time for it too form.

And since 0.3 number was CA choice so be it (for a fix, not mod).

[Bob the Insane]

Has anyone suggested yet doubling the threshold values for all the traits gained through combat to counter the apparant doubling effect that occurs when you fight battles on the battlemap??

[therother]

Has anyone suggested yet doubling the threshold values for all the traits gained through combat to counter the apparant doubling effect that occurs when you fight battles on the battlemap??

Yeah, but the problem is that the AI doesn't often fight on the battlemap. So if you doubled all the thresholds, it would take the AI twice as long to get each trait, which is no better than not doubling it (where the human player take half the time to progress up the levels).

One possible workaround would be to create a line of doubled battlefield traits that only the human player uses.

[Simetrical]

Except that you'd have to figure out a way to only have the player use them, which is tricky. There aren't any strat-map conditions that can limit a trigger to a player, unfortunately. You'd have to do it faction by faction, changing it for every game. Also, it would get screwed up if the player auto-resolved or the comp fought on the battle map, but both of those scenarios favor the AI, so that's okay.

How to work around this . . . you might be able to make some kind of conditional statement. Hmm. Wait, maybe I have something. What about this? Double all relevant thresholds, so my the last level of my example trait (GoodAttacker) has a threshold of 32. Then add a trigger (actually, two triggers, in case there's a limit on condition number) like so:

Trigger EvenGoodAttacker1
    WhenToTest PostBattle

    Condition IsGeneral
          and Trait GoodAttacker = 1
          or Trait GoodAttacker = 3
          or Trait GoodAttacker = 5
          or Trait GoodAttacker = 7
          or Trait GoodAttacker = 9
          or Trait GoodAttacker = 11
          or Trait GoodAttacker = 13
          or Trait GoodAttacker = 15

    Affects GoodAttacker  1  Chance  100

Trigger EvenGoodAttacker2
    WhenToTest PostBattle

    Condition IsGeneral
          and Trait GoodAttacker = 17
          or Trait GoodAttacker = 19
          or Trait GoodAttacker = 21
          or Trait GoodAttacker = 23
          or Trait GoodAttacker = 25
          or Trait GoodAttacker = 27
          or Trait GoodAttacker = 29
          or Trait GoodAttacker = 31

    Affects GoodAttacker  1  Chance  100

This solution isn't perfect. For it to work, you'd need to make all battle triggers have a 100% of giving you the trait(s) when their trigger conditions are met, which would make combat traits somewhat deterministic. To compensate, you could add selfperpetuating-type non-combat triggers to modify the traits up and down randomly, if you want.

Another problem is that battle triggers couldn't add any number of points other than one. AFAIK, we don't know how the triggers for a given event are parsed—if they're done sequentially, we could theoretically get around this problem by interspersing combat triggers with my new triggers. But it would probably be easier to just adjust all combat trait thresholds so that they work with two points per battle (which is what they'll get with the triggers set to give one).

Anyway, assuming that all battle triggers have a 100% chance of giving their traits, any general who fights exclusively on the battle map will always have an even number of points in his combat trait. If a general autoresolves, however, and no trait trigger gives a number of points other than one, he'll gain an odd number of points. The triggers above will immediately find combat traits with an odd number of points and bump them up by a point, effectively compensating for the lower payout of autoresolve.

Thoughts?

-Simetrical

[Bob the Insane]

Simetrical - sorry mate, you have complete lost me on that one...

I was thinking however that we could assume that the player largely fights on the battlemap and adjust the threshold to allow for that (doubling them)... We already know that the AI does not tend to gain high star rated general through fighting much, so we could try looking into a mechanisim to deliver high star rated generals outside of the battlefield...

Perhaps create a perpetual trigger for GoodCommand and BadCommander..

Theamin thing to consider is that anything we do will apply equally to player and AI..

[RJV]

Except that you'd have to figure out a way to only have the player use them, which is tricky. There aren't any strat-map conditions that can limit a trigger to a player, unfortunately. You'd have to do it faction by faction, changing it for every game.

But would that really be too much of a problem? Just add in the 'and not faction = whatever' condition, make n copies of the file, edit each to add the relevant faction name and voila. Couple this with a little batch file to rename the file from the faction-specific one to the genuine one before firing off the RTW exe and away you go. I'm skipping steps here of course, and you'd have to make sure you played the right faction once you loaded the game, so it's not all that elegant. But if it made it work....

Anyway, just something to think about.

Cheers,

Rob

[Sinner]

Simetrical, I get what you're thinking, but since the triggers are processed sequentially as far as I can tell I think we'll need something like this...

Trigger GoodAttacker1_31plus
WhenToTest PostBattle
Condition IsGeneral
and Trait GoodAttacker >= 31
Affects GoodAttacker 1 Chance 100

Trigger GoodAttacker1_30
WhenToTest PostBattle
Condition IsGeneral
and Trait GoodAttacker = 30
Affects GoodAttacker 1 Chance 100

Trigger GoodAttacker1_29
WhenToTest PostBattle
Condition IsGeneral
and Trait GoodAttacker = 29
Affects GoodAttacker 1 Chance 100

etc

This way, if I have 29 points in GoodAttacker, the only trigger I match is GoodAttacker1_29. It then increases my points to 30, but I don't trigger GoodAttacker1_30 since I'm already past it. There are a couple of potential problems though: as yet do we know if the game is scanning the list of triggers twice, in which case the second time around it will trigger GoodAttacker1_30 after all; the other possible flaw is that there might be a limit on the number of triggers, which we could very possibly hit given the number we'd need to have for each current trigger condition.

As for the faction specific idea, that could work albeit with a lot of effort to set up, with the easiest solution being to set up each faction as a mod and then running that faction specific version of the game using the -mod:XXXX switch in the command line options.

[therother]

As for the faction specific idea, that could work albeit with a lot of effort to set up, with the easiest solution being to set up each faction as a mod and then running that faction specific version of the game using the -mod:XXXX switch in the command line options.

Yeah, I've had much the same idea as Rob. The use of the mod is possible I guess, setting up a .bat script to run the game perhaps - press 1 to play Julii, and so on. I suppose that would be the least amount of work, in some respects, for the players and coders.

Still pondering the Simetrical/Sinner idea - has it been tested, or is anyone planning on testing it? So far, the best working solution that I've seen will just make the trait work as intended on average, and not for all instances. As I understand it, this would be a complete solution.

[SilverEye]

to get back on topic : the farmer/trader issue

Here's another 2 cents from from a fellow RTW player suffering from chronical poor farming at Frustration Level 1 Chance 100

reading through the entire 5 pages of this thread for the modding/scripting noob that i am and still trying to make sense out all of it i would like a conclusive answer to the issue.
Someone already said the same thing a few posts back without much result so it seems i will just have to give me the answer myself.

Well here goes: there is none...

I think it's safe to say that the trade/farm triggers are NOT bugged, but on the other hand saying that the default settings are a bit 'flawed' would be putting it mildly.
Well, with all the ideas, suggestions, testresults, modding examples etc given up until this point there is enough information here to play around with the trigger chances and trait thresholds to tweak and finetune it to your own preferences and playing style. Wich is exactly what i did, much to my own happiness.
I suggest that all of you who also cannot wait for the next CherryVanilla patch to adress this issue do the same, trust me you'll feel a lot better and no longer have that urge to throw R:TW out the window.

2 more things i want to say, then i'll shut up again.

First an example of what i mean with flawed default:
Aside from random births the only way to get BadFarmer is by building something other then a farm and the only way for BadTrader is by building something else then a trader, not building Ports and Roads do not affect BadTrader they only affect GoodTrader when you do build them , right ?
Then why is BadFarmer 1 chance 8% /building with trait thresholds of 1,2 and 3 and BadTrader only 1 chance 1% /building with thresholds of 1,3 and 6 ?!?

The default system is setup in such a way that only those few factions with an Agriculture or Trade Deity in their Pantheon will have a chance of avoiding the negative traits other then keeping their family member out of cities for their entire life, wich the other factions are condemned to.
Sure you may get 1 guy somewhere to go beyond the first level of the positive side of the trades with some absurd insane micromanagement, if he doesnt die of old age first before reaching NoGoingBackLevel 3. But even then there is no SelfPerpetuating or even a Genetics trigger to pass the hard earned virtue on to the next generation.
Specially in the later parts of the game you'll eventually run out of new farms to build since they can never be destroyed in newly conqured lands and you will end up with all your governors being Clones with the all the same negative Vices instead of unique characters with depth and personality no matter what you do...

thats what i mean with 'flawed' , its not bugged but it **** well needs fixing.
So either wait for a mod or patch or do it yourself.

Second ...
This is for the CherryVanilla mod people , both creators and users. Not sure if this has been adressed already somewhere else but i'll mention it again. After reading through this entire thread word for word and gaining some understanding of the Governing6 and Governing10 triggers and some modding of myself i noticed the following changes that were made:

Trigger governing6
WhenToTest GovernorBuildingCompleted

Condition not SettlementBuildingFinished > = farms

Affects BadFarmer 1 Chance 8
Affects BadAdministrator 1 Chance 10

Mod218 - 11/14 : added bad admin effect

Trigger governing10
WhenToTest GovernorBuildingCompleted

Condition not SettlementBuildingFinished > = trader

Affects BadTrader 1 Chance 1
Affects BadAdministrator 1 Chance 10

Mod221 - 11/14 : added bad admin effect

Now correct me if i am wrong, but does this not mean that building any trader or farm will give my governor a 10% BadAdmin chance and any other type of building a 20% BadAdmin chance ?!?

Ok that was all , i'l crawl back under my rock now.

[Sinner]

You are correct with your last statement, SilverEye. If that's CherryVanilla then it hasn't addressed the fundamental flaw with the BadFarmer trigger, if anything it's made the situation worse for governors in that they'd get BadAdministrator as well.

I personally consider the trigger condition for BadFarmer to be bugged in that it occurs even if you've already built the farm for that level of settlement, so it's impossible to avoid, but that's just my opinion.

I think the the most palatable answers so far are to either add the AdviseBuild condition since at least that takes into account whether that settlement level's farm is available to be queued or not, or to use SettlementBuildingExists to check if the farm has already been built for that level. Either method requires the single current trigger to be replaced by 5, one for each level of farm, and both need the % chance of getting BadFarmer points increased else the trigger becomes nearly ineffective.


Switching back to the battle triggers problem: from my testing it appears that they only get tested once for autocalced battles, but twice for tactical battles. I think there's a clue for what's happening in the code in the Events, Conditions, & Commands thread, if you look at many of the battle conditions, they can occur either in Battle or Strat (ie. Campaign map). I think what happens is that autocalc battles cause the triggers to be checked by the Strat game engine, while player-fought battles first get checked by the Battle game engine at the end of the combat and then again by the Strat game engine upon return to the campaign map.

Whether this is what is occuring or not doesn't really matter, since the effects are what's important for us and the simple and sad fact is that we're getting two trigger checks for every player-fought battle.

As a note, my testing method was as follows: I played as the Julii, first editting Vibius Julius to remove his GoodCommander 1 trait. I editted the V0090_Standard_Battle_Any_Victory_VnV_Trigger trigger to remove the BattleOdds condition since I wanted it to return a result simply upon victory, so the trigger looks like this...


Trigger V0090_Standard_Battle_Any_Victory_VnV_Trigger

WhenToTest PostBattle

Condition IsGeneral
and WonBattle

Affects GoodCommander 1 Chance 100


The only other triggers that give GoodCommander are those for adoption, random birth and worthy of marriage, so I know that no other trigger will be effecting my tests since I'm using an already established family member.

As I've already said, on autocalc victories Vibius gained only 1 point towards GoodCommander as would be expected, but for battles I fought he gained 2 points per victory.

I then tested the idea initially proposed by Simetrical although I used my slant on it, replacing the standard V0090_Standard_Battle_Any_Victory_VnV_Trigger trigger with the following...

Trigger V0090_Standard_Battle_Any_Victory_VnV_Trigger_8plus
WhenToTest PostBattle

Condition IsGeneral
and WonBattle
and Trait GoodCommander >= 8

Affects GoodCommander 1 Chance 100

;------------------------------------------
Trigger V0090_Standard_Battle_Any_Victory_VnV_Trigger_7
WhenToTest PostBattle

Condition IsGeneral
and WonBattle
and Trait GoodCommander = 7

Affects GoodCommander 1 Chance 100

;------------------------------------------
Trigger V0090_Standard_Battle_Any_Victory_VnV_Trigger_6
WhenToTest PostBattle

Condition IsGeneral
and WonBattle
and Trait GoodCommander = 6

Affects GoodCommander 1 Chance 100

;------------------------------------------
Trigger V0090_Standard_Battle_Any_Victory_VnV_Trigger_5
WhenToTest PostBattle

Condition IsGeneral
and WonBattle
and Trait GoodCommander = 5

Affects GoodCommander 1 Chance 100

;------------------------------------------
Trigger V0090_Standard_Battle_Any_Victory_VnV_Trigger_4
WhenToTest PostBattle

Condition IsGeneral
and WonBattle
and Trait GoodCommander = 4

Affects GoodCommander 1 Chance 100

;------------------------------------------
Trigger V0090_Standard_Battle_Any_Victory_VnV_Trigger_3
WhenToTest PostBattle

Condition IsGeneral
and WonBattle
and Trait GoodCommander = 3

Affects GoodCommander 1 Chance 100

;------------------------------------------
Trigger V0090_Standard_Battle_Any_Victory_VnV_Trigger_2
WhenToTest PostBattle

Condition IsGeneral
and WonBattle
and Trait GoodCommander = 2

Affects GoodCommander 1 Chance 100

;------------------------------------------
Trigger V0090_Standard_Battle_Any_Victory_VnV_Trigger_1less
WhenToTest PostBattle

Condition IsGeneral
and WonBattle
and Trait GoodCommander < = 1

Affects GoodCommander 1 Chance 100

I only went up to 8 points since I expected significant results long before I'd get any higher. I ran my tests again and for each autocalc victory Vibius gained a point towards GoodCommander, and yet again for each player fought battle he gained two points per victory, so this method will not work either.

So what answer is left? I think that leaves only the faction-specific option even though it's a major pain in the ass. It would also kind of force the player into fighting out every battle since he wouldn't get double points for autocalcing.

[therother]

So what answer is left? I think that leaves only the faction-specific option even though it's a major pain in the ass. It would also kind of force the player into fighting out every battle since he wouldn't get double points for autocalcing.

I'm beginning to think that there is no way around using player faction specific traits. Have been doing some tests of my own. However, things were delayed a little by me using Sobriety as one of the traits I used to test the triggers. With a British General. Yes, I know now...

[therother]

I'm beginning to think that there is no way around using player faction specific traits.

Okay, got the GoodCommander working properly using this idea.

I've made a new trait line, PlayerGoodCommander, as below:

;------------------------------------------
Trait PlayerGoodCommander
    Characters family
    AntiTraits PlayerBadCommander

    Level Confident_Commander
        Description Confident_Commander_desc
        EffectsDescription Confident_Commander_effects_desc
        GainMessage Confident_Commander_gain_desc
        Threshold  2 

        Effect Command  1 

    Level Good_Commander
        Description Good_Commander_desc
        EffectsDescription Good_Commander_effects_desc
        GainMessage Good_Commander_gain_desc
        Threshold  4 

        Effect Command  2 

    Level Superior_Commander
        Description Superior_Commander_desc
        EffectsDescription Superior_Commander_effects_desc
        GainMessage Superior_Commander_gain_desc
        Threshold  8 

        Effect Command  3 

    Level Great_Commander
        Description Great_Commander_desc
        EffectsDescription Great_Commander_effects_desc
        GainMessage Great_Commander_gain_desc
        Threshold  16 

        Effect Command  4 

    Level Legendary_Commander
        Description Legendary_Commander_desc
        EffectsDescription Legendary_Commander_effects_desc
        GainMessage Legendary_Commander_gain_desc
        Epithet Legendary_Commander_epithet_desc
        Threshold  32 

        Effect Command  5

I've also added/changed both these triggers:

;------------------------------------------
Trigger V0090_Standard_Battle_Any_Victory_VnV_Trigger
    WhenToTest PostBattle

    Condition IsGeneral
          and WonBattle
          and not FactionType britons

    Affects GoodCommander  1  Chance  100 

;------------------------------------------
Trigger V0090_Standard_Battle_Any_Victory_VnV_Trigger_Player
    WhenToTest PostBattle

    Condition IsGeneral
          and WonBattle
          and FactionType britons

    Affects PlayerGoodCommander 1  Chance  100

To change faction, all you'd need to do is replace "britons" with whatever faction you intend to use.

You would, of course, need to do this with every trait line which is affected by the battlemap (GoodAttacker, BattleScarred, BadCommander, etc.). Hmm, it's not a great solution by any stretch of the imagination, but at least it works.

(Quote the post to get the correct formatting.)

[player1]

Ok is this the way how Scarred and Coward traits and
Shieldbearer and Heroic Saviour ancillaries should work:

in export_descr_character_traits.txt:

Trigger battle1
    WhenToTest PostBattle

    Condition GeneralHPLostRatioinBattle < 0.7 ;was > 0.3

    Affects BattleScarred  1  Chance  30 
    Affects Brave  1  Chance  15 

;------------------------------------------
Trigger battle1b
    WhenToTest PostBattle

    Condition GeneralHPLostRatioinBattle < 0.7 ;was > 0.3
          and Trait Berserker >= 1

    Affects Berserker  1  Chance  30 

;------------------------------------------
Trigger battle1R
    WhenToTest PostBattle

    Condition GeneralHPLostRatioinBattle < 0.7 ;was > 0.3
          and CultureType roman

;    Affects BattleScarred  1  Chance  30 
    Affects RomanHero  1  Chance  15

and:

Trigger battle4
    WhenToTest PostBattle

    Condition GeneralNumKillsInBattle = 0 ;was not GeneralFoughtInCombat
          and PercentageEnemyKilled > 0 ;was = 0

    Affects Coward  1  Chance  10

And in export_descr_ancillaries.txt:

Trigger trigger_heroic_saviour
    WhenToTest PostBattle
    Condition GeneralHPLostRatioinBattle = 80
          and not Routs
          and WonBattle
          and GeneralFoughtInCombat
          and IsGeneral

    AcquireAncillary heroic_saviour chance  50

and:

Trigger trigger_shieldbearer
    WhenToTest PostBattle
    Condition GeneralHPLostRatioinBattle >= 0.2 ;was  5
          and WonBattle
          and not Routs
          and IsGeneral

    AcquireAncillary shieldbearer chance  30

[player1]

Disregard last post, < = got truncated

Ok is this the way how Scarred and Coward traits and
Shieldbearer and Heroic Saviour ancillaries should work:

in export_descr_character_traits.txt:

Trigger battle1
    WhenToTest PostBattle

    Condition GeneralHPLostRatioinBattle < 0.7 ;was > 0.3

    Affects BattleScarred  1  Chance  30 
    Affects Brave  1  Chance  15 

;------------------------------------------
Trigger battle1b
    WhenToTest PostBattle

    Condition GeneralHPLostRatioinBattle < 0.7 ;was > 0.3
          and Trait Berserker >= 1

    Affects Berserker  1  Chance  30 

;------------------------------------------
Trigger battle1R
    WhenToTest PostBattle

    Condition GeneralHPLostRatioinBattle < 0.7 ;was > 0.3
          and CultureType roman

;    Affects BattleScarred  1  Chance  30 
    Affects RomanHero  1  Chance  15

and:

Trigger battle4
    WhenToTest PostBattle

    Condition GeneralNumKillsInBattle = 0 ;was not GeneralFoughtInCombat
          and PercentageEnemyKilled > 0 ;was = 0

    Affects Coward  1  Chance  10

And in export_descr_ancillaries.txt:

Trigger trigger_heroic_saviour
    WhenToTest PostBattle
    Condition GeneralHPLostRatioinBattle < = 0.2 ;was >= 80
          and not Routs
          and WonBattle
          and GeneralFoughtInCombat
          and IsGeneral

    AcquireAncillary heroic_saviour chance  50

and:

Trigger trigger_shieldbearer
    WhenToTest PostBattle
    Condition GeneralHPLostRatioinBattle >= 0.2 ;was  5
          and WonBattle
          and not Routs
          and IsGeneral

    AcquireAncillary shieldbearer chance  30

[player1]

I just realized that GeneralFoughtInCombat is always true, even if general in not actually fighting, but just standing in the battle (that why I thought it was bugged).

But if you withraw general from battle then it will be zero.

So original trigger would trigger coward, if you retreat all your units at start without any fighting. Also, if enemy does the same instead.