feature: updates to blizzard sorc attack loop #707

jasorello · 2025-02-27T02:27:42Z

I had accidentally polluted my commits in #695 with bad line endings, this is a cleaned PR with just the blizzard sorc changes

This PR attempts:

Better static field usage, porting the more complicated logic over from the nova sorceress character, as well as moving the distance logic to the common config so it can be more easily referenced across all other sorceress builds
Moved blizzard-specific configurations to a function within the struct, keeping the same ease of configuration but without polluting the rest of the character package with consts (i also fixed some other cross-character references that were caught up in this)
I did some optimization of the blizzard attack loop. The problem I'm trying to fix here is that on a faster machine, we're iterating through this loop much faster than the bot can actually perform any of the attacks. We can hit 200+ attack loops on a boss after just performing 6 or 7 actual attacks within the game. I think the issues here may also be present in other characters, but I'll wait for someone smarter than me to weigh in.

I also included the switch from sorceress to blizzardsorceress that was suggested in #691

…onst; optimization of attackloop to minimize wasted iterations

jasorello · 2025-02-27T02:28:15Z

config/template/config.yaml

  useMerc: true
  stashToShared: false
  useTeleport: true # If set to false, bot will not use teleport skill and will walk to the destination

+  sorceress: # generic sorceress configurations


common configuration that can be leveraged via s.Data.CharacterCfg.Character.Sorceress.StaticFieldMinDist

jasorello · 2025-02-27T02:29:17Z

internal/character/blizzard_sorceress.go

 type BlizzardSorceress struct {
 	BaseCharacter
 }

+func (s BlizzardSorceress) attackConfig() map[string]int {


I noticed there were issues in the builds with accidental references across the characters which might have unintended results. I thought this might be a better pattern to make it more clear which settings are relevant for each character. Is there a better way?

jasorello · 2025-02-27T02:29:31Z

internal/character/blizzard_sorceress.go

@@ -53,6 +58,46 @@ func (s BlizzardSorceress) CheckKeyBindings() []skill.ID {
 	return missingKeybindings
 }

+func (s BlizzardSorceress) killMonsterWithStatic(bossID npc.ID, monsterType data.MonsterType) error {


This is mostly stolen from the nova sorceress

jasorello · 2025-02-27T02:29:49Z

internal/character/blizzard_sorceress.go

+		// Pull target threshold from config based on difficulty
+		var targetThreshold int
+		switch s.Data.CharacterCfg.Game.Difficulty {
+		case difficulty.Normal:


These values have a bit of a buffer over the bare minimum for the difficulty to minimize wasted casts

jasorello · 2025-02-27T02:30:32Z

internal/character/blizzard_sorceress.go

@@ -77,7 +128,8 @@ func (s BlizzardSorceress) KillMonsterSequence(
 			return nil
 		}

-		if completedAttackLoops >= sorceressMaxAttacksLoop {
+		if completedAttackLoops >= s.attackConfig()["maxAttacksLoop"] {
+			s.Logger.Error("Exceeded MaxAttacksLoop", slog.String("completedAttackLoops", fmt.Sprintf("%v", completedAttackLoops)))


I've noticed a few people in Help noting that their character is bailing early but not knowing why. I think many of them are from hitting this loop in their respective character. This log message should help identify those cases better.

jasorello · 2025-02-27T02:30:57Z

internal/character/blizzard_sorceress.go

@@ -86,39 +138,34 @@ func (s BlizzardSorceress) KillMonsterSequence(
 			s.Logger.Info("Monster not found", slog.String("monster", fmt.Sprintf("%v", monster)))
 			return nil
 		}
+		if monster.Stats[stat.Life] <= 0 {


This avoids unnecessary attackloop iterations for bosses with an extended death animation like andariel

Hello . Thats a good catch. There isnt any character using this yet but maybe it could be useful in that case?

could use npcmode 0 and 12 instead of life stat to determine if its dead if its problematic.

Yes I've been testing with this after every attack in KillMonsterSequence, seems to work:

if s.Data.PlayerUnit.Mode == mode.CastingSkill { attackSuccess = true }

Then at the bottom:

if attackSuccess { s.Logger.Debug("Completed attack loop", slog.Int("completedAttackLoops", completedAttackLoops)) completedAttackLoops++ }

I meant using mode dead and death . Dead is when it just died, death is when its extremely dead ( after animation) or i inverted

Yeah I meant to reply to the comment above about character bailing early, sorry

implemented, thanks!

Still have to check wich of the 2 mode is set after the animation otherwize it will still exit game early and have no real benefit from checking life stat

one of the mode tells its dead so we can stop attack sequence, the second one tells we can return and go on with next run( drops should be on ground/animation finished)

ahh gotcha i missed that distinction. i'll think about how I can implement that, i don't think it fits with how I set up the methods at this time, right?

Also keep in mind koolo is able to read itemlocation dropping . Both location ground and dropping are merged together. This means it reads item before they are visible on ground. Could explain early exit symptoms if nothing match .nip rule . We dont have to see them on ground to evaluate if it should pickup

I've been thinking about what you said and I'm not sure the distinction matters for the attack loop. Either the monster isn't dead and we should kill it, or it's dead and we should do something else.

Can you give some more background on the problem you're describing? Should there be a short wait added in the run definition or in the KillMephisto definition?

jasorello · 2025-02-27T02:34:17Z

internal/character/blizzard_sorceress.go

+			return nil
+		}
+
+		for s.Data.PlayerUnit.States.HasState(state.Cooldown) {


switched this to cast our leftskill until we're out of cooldown instead of just a single time. This should better leverage our activity during blizzard cooldown without resulting in a new attack loop.

I've pulled it ahead of the blizzard self-cast because otherwise we were usually hitting the self-cast block while blizzard was on cooldown.

New behavior should be:

while on cooldown: cast leftskill if timeToCastSelfBlizzard: cast selfBlizzard else cast regularBlizzard:

jasorello · 2025-02-27T02:34:32Z

internal/character/blizzard_sorceress.go

 		previousUnitID = int(id)
 	}
 }

-func (s BlizzardSorceress) killMonster(npc npc.ID, t data.MonsterType) error {


This function was no longer being called

jasorello · 2025-02-27T17:39:28Z

The blizzard logic is busted in this, that's what I get for doing a refactor late and night and pushing without testing. I've fixed it so it actually casts offensive blizzards again but I'm still seeing attack loops where nothing is performed.

e: I think i've resolved this

jasorello · 2025-02-27T18:05:45Z

I ended up pulling out the logic for 'is this monster dead?' into a separate function, i'll see if elobo91's suggestion can make it simpler.

…ifies things

…g NpcMode yet

updates to blizzard sorc; better static field usage; no more global c…

ce96a52

…onst; optimization of attackloop to minimize wasted iterations

jasorello commented Feb 27, 2025

View reviewed changes

jasorello added 2 commits February 27, 2025 12:46

make it actually cast blizzard again, add more debug statements

0659e45

more checks for dead monsters because they can die mid-attack loop

48e224c

jasorello added 4 commits February 27, 2025 13:10

somehow i missed that it was double-casting the leftskill, this simpl…

4d03f90

…ifies things

forgot i updated these sf options

2031bf2

Merge branch 'main' into blizzard_sorc_updates

6867046

isMonsterStillAlive pulled to BaseCharacter struct, but not leveragin…

af33116

…g NpcMode yet

jasorello force-pushed the blizzard_sorc_updates branch from 9fac835 to af33116 Compare February 28, 2025 02:33

jasorello added 2 commits February 27, 2025 22:23

switch to inspecting npcMode instead of npc life stat

c2e66c1

this method is actually generic, not unique specific (i think)

cf0bfa5

jasorello changed the title ~~feature: updates to blizzard sorc~~ feature: updates to blizzard sorc attack loop Feb 28, 2025

jasorello added 2 commits March 1, 2025 22:03

this was too much noise

6d09954

oops this isnt supposed ot be here

5fd5711

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feature: updates to blizzard sorc attack loop #707

feature: updates to blizzard sorc attack loop #707

jasorello commented Feb 27, 2025 •

edited

Loading

jasorello Feb 27, 2025

jasorello Feb 27, 2025

jasorello Feb 27, 2025

jasorello Feb 27, 2025

jasorello Feb 27, 2025

jasorello Feb 27, 2025

elobo91 Feb 27, 2025 •

edited

Loading

CarlPoppa1 Feb 27, 2025

elobo91 Feb 27, 2025 •

edited

Loading

CarlPoppa1 Feb 27, 2025

jasorello Feb 28, 2025

elobo91 Feb 28, 2025 •

edited

Loading

jasorello Feb 28, 2025

elobo91 Feb 28, 2025

jasorello Feb 28, 2025

jasorello Feb 27, 2025

jasorello Feb 27, 2025

jasorello commented Feb 27, 2025 •

edited

Loading

jasorello commented Feb 27, 2025

feature: updates to blizzard sorc attack loop #707

Are you sure you want to change the base?

feature: updates to blizzard sorc attack loop #707

Conversation

jasorello commented Feb 27, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

elobo91 Feb 27, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

elobo91 Feb 27, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

elobo91 Feb 28, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jasorello commented Feb 27, 2025 • edited Loading

jasorello commented Feb 27, 2025

jasorello commented Feb 27, 2025 •

edited

Loading

elobo91 Feb 27, 2025 •

edited

Loading

elobo91 Feb 27, 2025 •

edited

Loading

elobo91 Feb 28, 2025 •

edited

Loading

jasorello commented Feb 27, 2025 •

edited

Loading