#tc-research

1 messages · Page 45 of 1

stone light
#

What's the logic here?
What is it that the first condition doesn't catch that the second one does?
Why is the TTD inflated by 600 if they have the dot already?
I'm confused PepeConfused

unknown.png
twilit granite
#

Not sure about the 600 thing

#

I think the APL is making sure it spreads sbs properly for the max sim duration (600 sec)

stone light
twilit granite
#

The first call is applying it to the main target, the second one is cycling

finite sparrow
#

target_if=min:target.time_to_die+(dot.serrated_bone_spike_dot.ticking*600)
looks like this is used to ensure the TTD call is still prioritizing targets that do not have the dot. targets that do have the dot have a massively inflated TTD so they wont get picked until all targets have the dot

#

600s is probably a catch-all guarantee that this will happen no matter what custom raid events you are simming, but it does not need to be nearly this long for something like dslice specifically

knotty oriole
#

It's just a bit of a hack to add a level of filtering we don't have easy control over in the target_if criteria (if= is evaluated afterwards, we want to filter before a target is returned)

knotty oriole
#

Based on spot checking with current builds, going with this for reroll logic for now:

#

rtb_reroll,value=rtb_buffs<2&(!buff.broadside.up&(!runeforge.concealed_blunderbuss|!buff.skull_and_crossbones.up)&(!runeforge.invigorating_shadowdust|!buff.true_bearing.up))|rtb_buffs=2&buff.buried_treasure.up&buff.grand_melee.up

#

TB good to keep with Shadowdust, SnC good to keep with Blunderbuss, Broadside the only universally kept single buff

#

And BT+GM is also rerolled as Fuu suggested

#

0.2-0.3% depending on the build, so nothing too major here but seems to check out for now

hushed shell
#

Don't some of these look wrong? gm_tb is 16919 while a single tb is 17006. 87 dps / 0.51% difference doesn't seem like too high for it to be error margin, and it doesn't make sense why a single roll is higher than double, since in neither case the APL rerolled them, or am I missing something?

regal agate
#

error of margin is low because it is run at 100k iterations (so ~5 dps)
The buffs are the ones the sim keeps.
e.g. this is gm_tb:

unknown.png
#

this is TB

unknown.png
hushed shell
#

Yeah, so how come keeping a single tb is higher than keeping a tb+gm?

regal agate
#

you keep both GM and TB single roll

#

GM is a quite bad roll, so by keeping it it means you also have lower chance to roll better buffs

#

so the answer to "why is it worse to keep both TB and GM single roll compared to only TB?"
Because you keep GM

warm rain
#

I think the confusion here is that @hushed shell thinks gm_tb means keep gm tb double buff, while what it actually means is keep either single buff

regal agate
#

i did change the text to make it more obvious

hushed shell
#

I don't think you keep either single buff, not at least with the APL provided. It's an and clause & inside the parenthesis instead of an or clause with |

#

Is it not?

warm rain
#

the expression evaluates to 0 if either buff is up = dont' reroll

hushed shell
regal agate
#

it is actually de'morgans law, hence the negation
¬(A∨B) => ¬A∧¬B

warm rain
#

yes, the brackets are false, so it becomes true & false = false

regal agate
#

it needs to be true to reroll

hushed shell
#

Yeah, if either is false it's false. So it's never rerolling a single roll. A single GM returns TRUE & (FALSE & TRUE), which translates into TRUE & FALSE, meaning it doesn't reroll.

warm rain
#

i mean, that's the point

hushed shell
#

But this wasn't my issue in the first place, was just clearing out what Mirage suggested about the possible confusion.

#

My point is that if it's not rerolling a single TB, and also not rerolling a double TB + GM, wouldn't a double buff always be a head of a single roll, no matter what the secondary roll is?

regal agate
#

The lables is what it keeps

#

not what it rerolls

hushed shell
#

Yeah, that's what I mean. so it keeping a single TB is 0.5% increase of keeping TB_GM

warm rain
#

the line that does what you just said is the tb line

hushed shell
#

But how is keeping 2 buffs lower than a single

warm rain
#

both gm_tb and tb keep the gm+tb double buff, as well as the tb single buff

#

the difference is that gm_tb also keeps gm single buff

hushed shell
warm rain
#

we agree that a single gm roll means that the expression evaluates to false, right?

hushed shell
#

Correct, so it wouldn't reroll since it needs true to reroll.

warm rain
#

right... then that means it keeps gm single buff

hushed shell
#

Oh wait, my head is completely backwards

regal agate
#

rtb_buffs<2 basically filters it to be only single rolls
and !buff.x.up basically filters it to the individual buff

So example:
2 buffs are up:
rtb_buffs<2 is false so the entire condition is false, means no reroll
TB or GM is up:
!buff.tb.up&!buff.gm.up Tb or GM is false, means no reroll
RP or any of the other single buffs (that isn't gm/tb) is up:
!buff.tb.up&!buff.gm.up both are true, hence the condition is true, means it rerolls

hushed shell
#

Okay I get the gist now, I'm not sure how I was comprehending it but it makes sense now. It's basically trying to fish for those buffs, and rerolling if they're not there. Which makes sense why gm_tb is lower than a single tb since it's also rerolling more good buffs and keeping gm on top. Also makes sense why when snc solo -0.1% and solo rp -0.22% ends up together being rp_snc -0.11% since it's equalizing between the two. I think I was missing the fishing for the buffs part or something, not sure, but I was definitely thinking backwards for the moment.

Anyway thanks for sticking with me and clearing it up, sorry for the confusion and being somewhat dense.

finite sparrow
# 
actions+=/variable,name=finish_condition,op=reset,if=cooldown.between_the_eyes.ready&effective_combo_points<5```
this action in the outlaw apl is conflicting with other parts of the apl
#

since we dont cast BtE on cooldown anymore, this is forcing 5cp for basically no reason pretty often

#

removing it doesnt change dps much, but i noticed it when trying different finisher rules, only for them to be overwritten by this

knotty oriole
finite sparrow
#

makes sense i guess

#

i suspect this is only true for weaponmaster + blunderbuss + maybe triple threat

#

maybe too small of a gain and too specific of a setup to be added to the default apl

knotty oriole
#

Interesting

#

Still maybe worth looking at

knotty oriole
#

Obviously it's pretty marginal, 0.3% difference for 6 swapped globals. So the difference per cast is very tiny.

#

But don't really see any problem with adding something potentially here

knotty oriole
finite sparrow
#

probably, was just editing what was already in the apl (deficit>regen+10 is default if you remove the cp rule)

knotty oriole
#

Yeah I think that is basically just to accommodate for combat potency average

#

Gonna change the ordering around to make this a bit easier to analyze though

knotty oriole
#

It's -0.7% for normal build

finite sparrow
#

figures

knotty oriole
#

(Or pre-patch build that is)

#

Based on testing some various combos I'm leaning towards

#

actions.build+="/pistol_shot,if=buff.opportunity.up&(energy.deficit>energy.regen*1.5|!talent.weaponmaster&combo_points.deficit<=1+buff.broadside.up|talent.quick_draw.enabled)"

#

Seems the same for the main profile you linked but is also neutral in the Dust build and such

mystic jolt
#

Probably a super niche thing, but does the apl hold vanish if you have a 5 buff and GM or BT are the missing buffs until after you reroll?

lean talon
knotty oriole
#

Alright, integrated most of the pending APL changes as well as some random ones I was looking at and regenerated the default profile for visibility. Still working on the Blade Rush AoE logic, but everything else should be included.

# 
* Update RtB reroll priority for Blunderbuss and Shadowdust
* Update finish CP condition when using Blunderbuss
* Update Pistol Shot filler logic for Weaponmaster builds
* Add tab-Sinister Strike for Cache of Acquired Treasures bleed applications
* Add targeting logic for Flagellation to select the highest health target for multi-actor sims
* Update Blade Rush to not cast at high energy during Flagellation
* Update Between the Eyes logic for Greenskin's Wickers
finite sparrow
#

cycling targets with SS and axe is true at any target count? surprising

half karma
#

last I checked it stop being a gain at around 3 or more target but doesnt rly start becoming a loss

#

just a good gain at like 2 and 3 targets

#

and then its whatever

#

i guess its just because u cant really maintain the bleed on more than 2 or 3 targets so it doesnt really matter

regal agate
#

^ i cam to exactly this conclusion when doing the change back in march

knotty oriole
#

Other than it being annoying to do 😛

#

It's a pretty simple trade-off of crit chance from BtE debuff vs. the potential bleed application, which should be constant at any target number

finite sparrow
#

very random nitpick but i think pistol shot (_tornado_trigger) should be grouped under pistol shot and not main gauche

unknown.png
knotty oriole
#

Why is that?

#

It’s not part of the DPET of PS casts. It contributes to the DPE of MG procs

#

Grouping in the report is more about linked actions rather than similar spells

#

I might put BtE 4pc procs under PS though

finite sparrow
#

why should it contribute to the dpe of main gauche?

#

in details, the 2pc is counted as pistol shot dmg with the others

#

is that on details' end or from combat logs

mystic jolt
#

it'd contribute to main gauche's dpe because main gauche is what triggers it, even though the spell is a pistol shot
so without any "executions" of mg, the extra pistol shot damage wouldn't happen

warm rain
#

2t vent: https://www.raidbots.com/simbot/report/ap5CAV6SUSmVgdpm1p6X4w
2t kyr: https://www.raidbots.com/simbot/report/dcgKXekF2QnC3bE4vgDTVg
3t vent: https://www.raidbots.com/simbot/report/hWKBmoSetWwiAEtv15qtA5
3t kyr: https://www.raidbots.com/simbot/report/r1LcbKBP27Hma1k8kZHx3J
dslice vent: https://www.raidbots.com/simbot/report/ostcPfR9Lkk8MzWxjGCANv
dlisce kyr: https://www.raidbots.com/simbot/report/2t9pUrE4vppKL4BUZmfri9

i just copied the apl line from the outlaw one and added a new shadowstrike line, there's probably a better way of doing it

Raidbots
Advanced Sim - 1322131 Venthyr - 17,401 DPS - Raidbots
Raidbots
Advanced Sim - 1322131 Kyrian - 17,955 DPS - Raidbots
Raidbots
Advanced Sim - 1322131 Venthyr - 22,021 DPS - Raidbots
Raidbots
Advanced Sim - 1322131 Kyrian - 21,231 DPS - Raidbots
Raidbots
Advanced Sim - 1322131 Venthyr - 18,985 DPS - Raidbots
knotty oriole
#

Parenting it to PS would increase PS’s damage per execute which isn’t correct since it isn’t triggered at all by PS player actions.

knotty oriole
warm rain
#

Ah I see, tyty

knotty oriole
#

But Fuu does have a change for Sub lined up already. So it’ll be in soon.

knotty oriole
regal agate
#

Only spec it wasn't rly impactful to optimize for cache is for assassination

finite sparrow
#

trying out different vanish conditions for weaponmaster blunderbuss, the current rule is at ambush_condition only. I think they can be free to do it at ambush_condition or finish_condition. the logic being that SS is probably a better global than ambush

interesting enough, forcing finish_condition only does not seem to be a gain https://www.raidbots.com/simbot/report/chNeiETAcynbykNnsk2S5n/simc

unknown.png
knotty oriole
#

Also assume it might need to get moved inward a bit. Instead of:

vanish,if=!runeforge.mark_of_the_master_assassin&!runeforge.invigorating_shadowdust&!stealthed.all&(variable.ambush_condition&(!runeforge.deathly_shadows|buff.deathly_shadows.down&combo_points<=2)|talent.weaponmaster&variable.finish_condition)
Maybe this:
vanish,if=!runeforge.mark_of_the_master_assassin&!runeforge.invigorating_shadowdust&!stealthed.all&(variable.ambush_condition|talent.weaponmaster&variable.finish_condition)&(!runeforge.deathly_shadows|buff.deathly_shadows.down&combo_points<=2)
otherwise I think it'll mess up Deathly Shadows logic

finite sparrow
#

yea i tested it with several profiles from fuu's sheet and it was either neutral or a small gain for any weaponmaster setup

knotty oriole
#

Nice

#

Yeah just wasn't sure if it was a BB thing or a WM thing or both

finite sparrow
#

i wonder why only vanish > dispatch is such a loss though

knotty oriole
#

Still finding it mildly amusing that WM has been trash tier for Outlaw for ages then it's suddenly good. 😛

finite sparrow
#

reminds me of ghostly strike simming incorrectly for an entire expansion

knotty oriole
#

Well at least that actually was kinda explained

#

WM has just functionally not been good lol

finite sparrow
#

yea, last i remember it was considerable for weird triple Wits setups

knotty oriole
#

Think strictly constraining Vanish to Dispatch results in it munching some BtE casts

finite sparrow
#

that makes sense

knotty oriole
#

Only 13.8 hardcasts of BtE on that profile vs. 14.6 in the other profile

#

Which isn't a massive difference but like 1% less buff uptime would explain that difference

#

So maybe it could be worked around in some other way if also checking for some BtE condition or something but that might be overly complex without any gains

solar swift
#

is it possible to specify the initial weapon from the cache trinket? it would be useful for double-use trinkets sim

snow abyss
#

yea i feel like its a dps gain to hold fcm if you start with haste buff less then 8 seconds

#

and use axe first

regal agate
tender brook
#

A niche situation but something I've run into the last couple weeks: could someone sim if it is a gain to pistol shot into finish in the last 2 globals of flag if you have a 4set proc pooled but no opportunity proc? I've just been feelycrafting that it's a gain especially if you have a bad flag and are sub 30 stacks that the flag stacks+flag damage+cdr is > the loss from a naked pistol shot. I'm not good with the apl stuff or I'd sim it myself

finite sparrow
tender brook
#

that is what i figured, it is something that doesnt pop up every fight just seemed like in that situation it could be worth to use the global on pistol shot

#

Thanks for checking those sims though

lime canyon
#

I feel like it's something that may happen more often with unoptimal play, and it does seem to be a dps increase?

knotty oriole
#

It’s all within error margin in that report so it’s not really an increase.

half karma
#

also worth noting that this rule is even better when playing Dancing Steel in dg for exemple since u are probably more energy starved without playing BR

muted sorrel
#

Is ignoring ambush so that you get more finisher/fish for blunder

#

Or to save energy/both

viscid oracle
#

Is CTO treated like regular buff on simc

knotty oriole
#

Yes. Obviously they have different duration and are refreshed differently but they all map to the same buffs.

rare vale
#

It also has confusing interactions with reroll, but yeah same as regular buffs

knotty oriole
rare vale
#

Sure but sometimes they reroll, sometimes with low duration on RTB they don't

knotty oriole
half karma
#

idk if its implemented in sims

knotty oriole
#

CtO buffs are preserved after rerolling in general, as long as they don't conflict with the new roll

rare vale
#

Very frequently, rerolling results in loss of CTO buffs

half karma
#

they must have more duration than normals buffs

#

for them to be preserved

knotty oriole
#

I'd have to see some controlled version of this because as far as our last controlled testing showed, the only time the buffs weren't preserved was if the new reroll overwrote it. They used to reroll a new buff in this case but that functionality was removed in beta.

#

But there are some odd cases happening where the behavior is not correct, which could be a side-effect of all of this

#

Always possible some functionality was changed though considering how much messing around with Outlaw they have done with the set bonus and hotfixes and such

#

Would be straightforward enough to add the constraint you're describing. Don't think it would impact sims much though because the overlap window here is small and given the reroll timing in SL (with the ability CD how it is) and lack of aggressive rerolls, this likely would only be a small loss in buffs on average.

half karma
knotty oriole
#

Sims currently reroll CtO buffs away when the reroll buffs conflict with them. Not always. If there is a currently duration constraint then no, sims don't currently implement that as that wasn't the last known functionality when this was last tested in-depth.

Quick testing in-game does seem to indicate there may be something going on as you describe, although I found at least one case quickly that didn't conform (1 buff was dropped and 1 was maintained when they were both less than the reroll duration.) It does seem to be somewhat inconsistent, not to mention the reroll bug that has been described a few times where the full duration buff doesn't reroll.

#

I will note that testing this change in SimC makes virtually no difference though

#

To the end result, since it's so infrequent

half karma
#

I know that there is some strange thigns happening in others scenario

#

but this one is a 100%

knotty oriole
#

I mean I could believe it's the general rule but I dunno if it's 100% since I just observed it not happening on a double CtO into reroll in-game.

half karma
#

You probably didnt see it right, I play a lot and have a wa to see wich buff is cto or not and this behaviour was always consistent

#

but other than this case of course the rest is kinda unpredictable

knotty oriole
#

I just observed a 3s and a 9s CtO buff where only one went away on a ~20s RtB reroll. Generally it seems to hold what you are saying but I'm just saying I think there may still be some inconsistent behavior here.

#

Happy to make any changes if they are confirmed but testing this in SimC seems to make no notable difference fwiw

#

It's within error margin at 0.05% error

half karma
knotty oriole
#

Normal roll was higher than both CtO buffs yes

half karma
#

im saying that if cto > normal buff it will never reroll

knotty oriole
#

But one of the CtO buffs persisted for some reason instead of disappearing as expected

half karma
#

yeah the other way do some weirds things sometimes

#

like extending and such

knotty oriole
#

Pretty rare to have a long duration RtB with RtB CD available and 2 CtO buffs though

#

So it's hard to test this repeatedly

#

Since we store off the CtO buffs in SimC before restoring them after the reroll, this behavior is easy to update

#

e.g. right now the code is

    for ( buff_t* buff : buffs )
    {
      if ( save_remains && buff->check() && buff->remains() != remains() )
      {
        count_the_odds_states.push_back( { buff, buff->remains() } );
      }
#

To store off the CtO buffs for after the roll

#

Which then just becomes

#

if ( save_remains && buff->check() && buff->remains() > remains() )

#

Based on the rule you describe

#

But with the current APL this doesn't really yield any DPS change. Still worth changing if it's the correct behavior though.

#

These are the two sims

#

With the different versions

#

DPS Error margin of 7.6 / 0.049%

#

Makes sense that it's low impact. We don't usually have much opportunity to reroll at high duration base buff (outside of maybe Dreadblades or something where we're spam finishing or with TB up) the window for normal CtO buffs to be rerolled is quite small since the buffs are short, and any time it would matter (Stealth duration buff towards the CD coming up) it's got a decent chance of being higher duration anyway.

#

(Also if we have a useful CtO buff it still keeps us from rerolling right now in the APL)

half karma
#

well the situation happens a lot in real gameplay so a new rule could probably yield some more dps

knotty oriole
#

Well that would indicate to me that people aren't totally following the same reroll rules as in the APL?

half karma
#

well i am but the thing i told u earlier is just free to do

#

it can't be a dps loss

knotty oriole
#

Shouldn't be more or less common of a statistical occurrence

knotty oriole
# half karma well i am but the thing i told u earlier is just free to do

What are you suggesting would be different about the APL/gameplay as a result of this?
If we have a good CtO buff we already won't reroll, and we only non-reroll RtB when duration is <3s which dodges most of the potential overlap.

Based on the sim putting some debugging in, it looks like this happens on average 2.2 times per 5 minute sim. (Out of 33.7 CtO procs) Basically once every 5.5 RtB casts on average, but the average duration lost is very small even in these cases. And is skewed towards bad buffs.

half karma
#

right now the apl wouldnt reroll if the cto buff was good

#

in real scenario u can

knotty oriole
#

Well in theory with the previous functionality in SimC we could already do that (since it assumed we wouldn't normally lose the CtO buff) but I don't think we ever saw any gain from doing this. Since you still have a ~20% chance of losing the CtO buff due to reroll overlap.

#

(Or not losing the buff itself, but the value of the CtO proc)

#

APL could probably already be tested for this by checking the durations

#

To see if it is worth ignoring CtO durations in the keep conditions

#

(e.g. !(buff.broadside.up&buff.broadside.remains=buff.roll_the_bones.remains) or something)

knotty oriole
#

But basically you have a two main situations:
a) You have a good CtO (say Broadsides) buff on short duration, which will prevent you from rerolling for a couple seconds. Probably not major impact since it's sub-5s.
b) You have a good CtO buff on a longer duration, which will prevent you from rerolling. This is mostly where your suggestion would impact. Often your main buff will be sub-15s at this point just due to the structure of the CDs, so you could reroll and keep your Broadsides.

The question of b) is if it's worth the ~20% chance of throwing that duration out by natural rolling that buff again. Which throws away maybe 5-10s of Broadsides time you could have kept by delaying the reroll until later.

Probably hard to answer that without doing sim comparisons.

half karma
#

yeah i see what u mean I didnt think that you would have potentially a chance to "lose" the cto benefit

#

but you would also extend it right cause we have pandemic on rolls buff iirc

knotty oriole
#

afaik if you natural roll Broadsides in this case the CtO duration is discarded rather than added to the final result (e.g. if you are getting a 30s reroll you'll just get a 30s Broadsides buff, not pandemic.)

#

But definitely worth investigating

#

I think all the results of the RtB roll are influenced only by the calculated roll duration, not any existing CtO buffs

#

(The duration of the hidden container buff)

#

315508

#

I kinda suspect this case is also what may cause the refresh bugs

#

That it may be the case where a CtO buff that exists that then natural rolls is still being considered a CtO buff for the next reroll and then has odd refresh behavior

#

(Probably based on duration as you describe, but since it has a equal duration maybe it doesn't get refreshed. Could be >= in this case rather than just >)

#

Hard to say though, whole thing is super wonky 😄

knotty oriole
finite sparrow
#

what does movement actually do for melee dps sims?

#

e.g. the raid events of HAC

raid_events+=/movement,distance=25,first=22,cooldown=33,last=337
raid_events+=/movement,players_only=1,distance=8,first=13,cooldown=18```
knotty oriole
# finite sparrow what does `movement` actually do for melee dps sims?

It essentially makes them move 8 yards. Which incurs a time penalty for being out of range (based on current movement speed.) Players can still use things that are in range (e.g. Pistol Shot) that are usable while moving (not channeled things, no caster spells with cast time unless they have some way to bypass)

#

Technically we could add Shadowstep or something into our APL to bypass this

regal agate
#

if the actor moves to a higher distance, it will just act as a interrupt of actions more or less

knotty oriole
#

But there has never really been much of a reason to

#

It's an arbitrary restriction in the sim type after all

#

To simulate disruption I suppose

#

Shadowstep/Hook would only really be relevant for relative comparisons between specs and I'm not sure if HAC is appropriate for that anyway

regal agate
finite sparrow
#

if the distance is under 8y would it basically do nothing?

#

for melee

knotty oriole
#

Movement mechanics work fine for melee but generally aren't relevant for anything other than like DH and maybe Monk

finite sparrow
#

er well i guess only outlaw gets the 8y melee range

knotty oriole
#

DH does simulate brief distance movement from using FR/VR

#

To get back in hitbox range

#

But since there's not really any "tradeoff" here other than maybe exploring when it's worth using PS/Toss/PK when out of range

#

Not super relevant for Rogues or most specs in general

finite sparrow
#

a small gain for a rule that is almost certainly simpler than energy.deficit>energy.regen*1.5 (just moved ss to be higher prio than pistol shot)

livid crow
knotty oriole
#

I’m assuming this is also negative when running QD+BB? Not sure if you checked that.

#

Not at my PC right now but I can poke around later this afternoon.

finite sparrow
#

khar is right that its a small loss in dslice (not getting 1% tho)

#

yes, i assume its only a blunderbuss thing

livid crow
finite sparrow
#

i see

livid crow
#

I was also trying to test whether skipping opportunities where you proc tornado trigger was a gain or not but it's a push

#

because you're going to get tornado trigger eventually, I didn't see a reason to force it. there's probably latent gains from stuff like natural BTE uptime or not overcapping CPs when combining tornado trigger and BB

finite sparrow
#

the 4pc can only be consumed with a casted pistol shot

#

the 2pc wont keep cycling it

livid crow
#

with a BB shot?

finite sparrow
#

yea bb or just regular opp

livid crow
#

yeah, was assuming that you'd get the tornado trigger through BBs passively, but maybe it's also sucking because 2pc shots are getting wasted while you wait?

finite sparrow
#

yes, the 4pc would be sitting there when it potentially could have already been fired and started cycling again

#

which may still be worth less than spamming ss for bb

livid crow
finite sparrow
#

but idk

livid crow
#

notably, the prio damage from any of these other options seems to be like +2.5%, which doesn't really make sense to me if it's tying in ST

half karma
#

probably the fact that the apl is still using BR on cd in aoe, making it even less usefull to use opp proc than in st

livid crow
#

it is likely something relating to BR, but the -1% aoe rotations should all be using fewer opp procs. I did some surface level investigations earlier and it's remarkable how much nothing else changes. casting drastically different SS/PS and getting the same CP throughput, CTO, CDs, etc

half karma
#

well the one at the bottom should cast more opp proc

unknown.png
#

than the 3 others

livid crow
#

can you link to that report?

half karma
#

its urs

livid crow
#

oh that's prio target damage, yeah

#

okay that makes sense. I thought you meant overall damage in relation to BR

half karma
#

no yeah

#

u were talking about prio so

knotty oriole
#

Yeah I don't quite understand why the difference in prio damage there would be caused by those changes

#

It's worth noting that the Cache damage in the Outlaw Venthyr profile here is much higher which is interesting

#

(For the AoE sim)

finite sparrow
#

its cache

#

cause the way i made the apl change

#

i didnt change the prio of tab target ss

knotty oriole
#

Ah right so it's just changing the Cache to be focused on the ST instead of AoE spread which biases the prio calculations

finite sparrow
#

so the other configs just might not be doing that at all

knotty oriole
#

That makes sense looking at the other stuff

#

Everything else is lower or the same so must be Cache

finite sparrow
#

i new a lazy apl change would bite me dead

#

with the cache fix, dslice and cleave are similar gains as st for this profile

hidden gyro
# finite sparrow with blunderbuss, thoughts on this rule for opp usage? <https://www.raidbots.com...

if you are talking about this,i can share my friend's thought,he is the people who create this usage.He thought that it's a risky usage and it should be use when you have adr,Bloodlust and flag to proc more BB for more damage,but you might be unlucky to proc BB when u ss until you don't have enough energy so that u only can consume proc opp.when u in the normal situation,proc opp should be consume 60-ish(tested by other theorycrafter).

finite sparrow
#

that sounds very close to the default rule energy.deficit>energy.regen*1.5 which would adapt to having adr/lust active

hidden gyro
opaque sedge
#

the china dude actually thought it was same as what solo said,we had the chat

#

as in u send opp proc if u dont have energy for SS

finite sparrow
#

it was from ramfam telling me that chinese tc'ers were doing it

livid crow
#

I have like no experience in this specific topic but what's the advantage of sending a PS immediately instead of waiting for another SS? If SS is so great that you ignore PS in all other moments

opaque sedge
#

can recall ramfam saying something on stream as well,as in when u actually wanna send opp proc but couldnt rly understand what he was talking about.But it was mentioned that there was a reseasrch from a chinese guy earlier than us

finite sparrow
livid crow
#

if using opp procs was good because of energy efficiency, then it should be better to be using opp procs much earlier than as a last resort

#

so e.g.
"/pistol_shot,if=buff.opportunity.up&(energy.deficit>energy.regen*1.5|!talent.weaponmaster&combo_points.deficit<=1+buff.broadside.up|talent.quick_draw.enabled)"
to
"/pistol_shot,if=buff.opportunity.up&(!talent.weaponmaster&combo_points.deficit<=1+buff.broadside.up|talent.quick_draw.enabled)"
is what that would imply I think (specifically for BB, idk if that messes up other builds)

hidden gyro
finite sparrow
#

hmm opp is likely to proc combat potency so its very close to a neutral global

livid crow
#

going from nearly-zero to zero opp proc usages is probably well within the margin of error so idk

finite sparrow
livid crow
#

yeah never opp uses way less PS. not really "nearly-zero" I guess

#

my gut says there's a target number of PS you want and "opp only if you cant ss" rotation and normal rotation are sort of on either side, with a steep cliff

finite sparrow
#

oh also for this profile, using dancing steel causes the new opp rule to just be neutral

#

with blade rush the never opp is not as bad

unknown.png
livid crow
#

I'm pretty sure there's a haste concern involved also, since opp is good for energy

#

1.7% is a lot to overcome through stat swapping though

finite sparrow
hidden gyro
#

does the bb still proc ramdomly in sim?

#

bb are so uncertain that when using might cause the data unstable when simulating

#

it might cause the difference about the sim

finite sparrow
#

simc uses the same proc chance as in game

#

it's doing a pull like 30,000 times (or whatever your iterations are) and averaging the result

hidden gyro
#

oh i see

opaque sedge
finite sparrow
#

nah, its just combat potency can only proc from offhand autos, main gauche and pistol shot

#

and its a 75% chance

opaque sedge
#

ah i c,it makes sense

#

i know realized,even after so many years i didnt even know that

#

insane what u can learn

knotty oriole
#

But I'll take a look if anything is odd there

#

But generally I think it's still useful to use Opportunity and realistically the only time it's not is when it will cause you to cap energy. (Which is what the regen condition is for.) But it could be there is some more specific edge case that causes SS to be better which is what the 0.2% is eeking out

#

Might make sense to try to isolate specifically what that case is

#

Because there could just be a bigger increase to find with some more restrictive condition in when SS is preferred

#

I think it's clear Opportunity PS is better than SS the majority of the time based on the final sim there though

livid crow
#

I think it at least needs a haste rebalance before a conclusion can be drawn from the same gear set. there's a pretty big downtime difference between them

finite sparrow
#

it "makes sense" that pooling energy for ss is not a gain since pooling has not been a thing in the history of outlaw, but i couldnt necessarily put into words as to why

knotty oriole
#

In the no-Opp (only using it on BB procs) APL it's essentially wasting almost exactly 50% of the Opportunity procs

#

In the normal APL it only overwrites ~25% of the Opportunity procs

#

Opportunity Pistol Shot is extremely efficient so it should basically "always" be better as long as you are resource-constrained or the DPET of SS doesn't become too large

#

There are likely some cases the energy.time_to_max condition is not capturing or some other case (maybe with SnC or AR max energy expiry?) where it's not beneficial but is currently considered as such which is what the re-ordering is exposing

#

For example in your original sim @finite sparrow there's a 10 energy difference on average in reduced AR energy expiry which could be contributing to that 0.2% efficiency

#

92.97 Overflow on Adrenaline Rush (Expiry) vs. 83.44

finite sparrow
#

adr expiry?

#

is that when the +50 max energy is taken away?

knotty oriole
#

Yes the Overflow there is something I added to track how much unused Energy is lost

#

When AR expires

finite sparrow
#

i see

knotty oriole
#

I suspect this is an area we could improve for sure

#

And could be contributing to this

livid crow
#

the SS/PS number of uses difference on this latest sim barely exists, for the record

knotty oriole
#

So this is confirming my hunch a bit

#

This is me just changing the existing condition to be this

# 
energy.deficit>energy.regen*1.5&(!buff.adrenaline_rush.up|energy.deficit-50>energy.regen*1.5)
finite sparrow
#

'asd' is very descriptive 😛

#

oh interesting

knotty oriole
#

So I think we just need to make something a little smarter for handling AR energy deficit handling

livid crow
#

also in general doesn't this mean it's safer to err on the side of ignoring opp procs closer to the "opp only if you cant ss" instead of 70? e.g. rule of thumb can be 60/50/40 w.e

knotty oriole
#

(Could just change it to energy.deficit-(buff.adrenaline_rush.up*50) but might just put it in a variable)

finite sparrow
#

the 70 thing is just what i say to explain energy.deficit>energy.regen*1.5

livid crow
#

not sure what the dps penalty looks like when you increase the rule of thumb so that you're overcapping instead, but you're practically pushing/gaining by going lower

knotty oriole
#

Might only be worth considering if AR duration is <5s or something idk

#

Which would be the advantage of variablizing it

finite sparrow
#

i.e so yes the number 70 is not significant as it's just an energy threshold that is usually close to that deficit rule

#

so i tend to say that instead of telling people to start tracking their energy regen number

knotty oriole
#

That's fair

livid crow
#

right I get it's an abstraction, but if people are trying to hit 70 and accidentally overcapping, would it generally be better to recommend even lower, if these numbers are showing a push by going literally all the way down?

#

e.g. going higher is risk for no reward. not everyone is tracking their flag/AR/haste/combat potency etc perfectly in the global

finite sparrow
#

The numbers are showing a push only because energy.deficit>energy.regen*1.5 can be better tuned to adapt to the increased maximum energy of ADR

#

fixing this will not change 70 as an abstraction

livid crow
# knotty oriole

also this needs to have a 4th option where it has the ADR fix and the opp if no SS together right?

#

or does that even make sense. I think they might be mutex

knotty oriole
#

Doing something similar helps KS as well looks like

#

Since it also relies on deficit values

#

This is for KS instead of BR

#

BR probably needs to be converted to consider something like this too though since it's just using time_to_max.

#

Which is based on deficit under the hood

#

Probably won't keep this a variable, just going to do it for prototyping. I'll probably add an energy.base_time_to_max and energy.base_deficit expression extension in code

#

Ok this is for Blade Rush fixing the time_to_max to ignore AR overage

finite sparrow
knotty oriole
#

Compared to 93.43 in the base profile

#

Obviously won't be able to mitigate all of the overage here since there's just naturally gonna be some times we cap no matter what we try to do, especially during lust

#

But this is certainly more efficient

opaque sedge
#

Does that mean solo was actually right or we dont know yet

#

(on sending opp proc when our energy is not enough for a strike)

finite sparrow
#

no

opaque sedge
#

nvm the sim says so

#

its less

#

what about the ''variable''

finite sparrow
#

sap

opaque sedge
#

cant track on chat

#

okok,ill let u do ur thing,nvm me

finite sparrow
#

it means the expression energy.deficit>energy.regen*1.5 can be better adapted to not waste the 50 energy lost from adr falling off

#

so the rule is not actually changing

opaque sedge
#

ah nice

knotty oriole
#

Yeah variable in this case is something like this

# 
actions+="/variable,name=base_energy_deficit,value=energy.deficit-(buff.adrenaline_rush.up*50)"
actions+="/variable,name=base_energy_time_to_max,value=variable.base_energy_deficit%energy.regen"
#

But I'll be shifting this into code

finite sparrow
#

in other words simc just wasnt always doing what we I thought it was

knotty oriole
#

I mean it was doing what we thought it was

#

It's more that all this means is it's not worth playing around the AR max at all

#

It's better to just consider our max energy is the base energy all the time

#

Tested some various duration checks like <5s for example and those weren't any better

finite sparrow
#

so using 70 as an abstraction is even more correct than before

knotty oriole
#

(They were better than baseline but not better than just ignoring it entirely)

finite sparrow
#

instead of like, having to think of it changing to 120 or something while adr is active

knotty oriole
#

But yeah I'll move this into code so we'll have options of using either energy.deficit or energy.base_deficit as appropriate

#

With base_deficit ignoring any temporary max

opaque sedge
#

does that mean we were actually undersimming

finite sparrow
#

a lil

knotty oriole
#

Not really, it's just a different way of approaching things

livid crow
#

I think a scenario where you can use the ADR extra cap is like trying to BR or KS within the window and using the extra room to get away with it, then being more moderate afterwards to bring the energy back down. that would let you get your restless blades working sooner. probably too small to actually do anything though

knotty oriole
#

That could gain some more damage

#

Primarily around BR and KS seems like more than anything

#

Because they are more restricted to non-cap Energy scenarios

#

After all the gain to the PS rule is only like 0.1-0.2% of this, whereas the KS/BR conditions are gaining another 0.3-0.4% from ignoring the AR max

#

This is better than just binary ignoring KS and BR entirely during AR like we have tried in the past

#

Since there are plenty of cases where AR may be up but our energy is basically 0

#

And in that case it's still worth using

opaque sedge
#

i ant rly imagine having ar up and pressing KS

#

even on dead energy it feels so overkill

#

u go from 0 to overcap half way in

knotty oriole
#

I mean this is probably fairly relevant for Celerity for example. A straight AR restriction would be very limiting for no reason.

opaque sedge
#

yeah gotcha

knotty oriole
#

Or maybe early in AR duration if you're at low energy to start

#

But yeah

#

This will still consider AR's regen, just against the base max energy, not the adjusted max

#

In practical terms, all this finding means is "pretend your max energy is always base max energy and ignore the AR max because we can't reasonably dump our energy fast enough to not lose it most of the time"

livid crow
#

without this fix, wouldn't celerity be doing pretty dumb stuff with your SS/PS decisions when you get celerity?

knotty oriole
#

Possible, although since Celerity procs often will happen at low energy randomly it's probably fine most of the time anyway

#

But yeah it could certainly have funny cases

#

BR and KS timing is the most relevant here since both of those CDs are currently trying to only be used when energy starved. The PS vs. SS consideration is only a minor difference.

#

Since this also slightly reduces the overflow from BR energy gain itself in addition to the AR overflow reduction

#

Basically this is "generating" about 30 energy and swapping a few PSs for SSs when it wouldn't be beneficial. It's not huge but we'll take it. 😛

#

Since it's a rather simple change

finite sparrow
#

sorry @opaque sedge i see how what i was typing was misleading

opaque sedge
#

im trying to soak information,mostly buzzing around,creating issues from nowhere

finite sparrow
#

ultimately this is the tldr

unknown.png
#

even though i never really told people to think of 70 changing to a different threshold during adr anyway

#

and br/ks on st will remain 'use at low energy'

#

about to just put these energy rules into a weakaura and stop trying to reword them in a normal way dead

hidden gyro
#

guys do you get the conclusion?

finite sparrow
#

yeah, the difference was only from wasted energy when adrenaline rush falls off and removes its +50 max energy

#

the 'opp only if you cant ss' rule was better at dumping energy before this happens

half karma
# knotty oriole

what exactly is the final rule is in this sim? And why can't we just put "use ps if under a certain treshold of energy" instead?

knotty oriole
#

The rule is just ignore the additional max energy from AR when considering deficit/time_to_max

half karma
#

time_to_max only change if on AR or not right ?(and a bit depending on haste as well but could be ignored?)

finite sparrow
#

(ttd>10 is about .4%)

regal agate
#

>=16 is what i use for subtlety

finite sparrow
#

fair

#

seems about the same

#

20 is a small loss in HAC but 16 is neutral

knotty oriole
#

Similar to what I added for Flag

knotty oriole
#

Don’t think Sepsis will ever be cast on the adds in HAC

knotty oriole
#

Seems to indirectly force some alignment with Vendetta ending up overlapping with Chains trinket usage

#

e.g.

actions.build="sepsis,target_if=max:target.time_to_die,if=target.time_to_die>20

This is actually slightly "worse" (+0.4% instead of +0.8%) and the only difference is trinket damage, even though it should logically always be better from the standpoint of uptime. The "problem" here seems to be that it finds a valid target sooner.

#

It's quite odd though, I don't really see why it should matter

#

I need to look into the direct cause some more, very curious

knotty oriole
#

Guess BtE also needs to be up otherwise it's losing a fair bit of damage so need

max:target.time_to_die*debuff.between_the_eyes.up

#

Neither really has any impact on DungeonRoute either so seems specific to perhaps DS timing. But with a BtE conditionals it seems OK with 11 also

#

actions.build="sepsis,target_if=max:target.time_to_die*debuff.between_the_eyes.up,if=target.time_to_die>11&debuff.between_the_eyes.up|fight_remains<11"

finite sparrow
#

isnt bte like always up?

#

er i guess you can start a new pack with bte still on longish cd

knotty oriole
#

Yeah but the highest HP enemy isn’t always going to be your primary target in DS with overlapping waves

noble pendant
#

Some revisited logic on sin that was marginally worse in 9.1, clipping flag for a little bit of extra stacks seems to be worth it.

Single target: https://www.raidbots.com/simbot/report/8295jTVd1PFFJhTcVhg1Ek
Cleave Add: https://www.raidbots.com/simbot/report/odrUmRrQUGzF62qfN26fj3
2T: https://www.raidbots.com/simbot/report/evKkDCEZfMbSw22dret8Ex
HAC: https://www.raidbots.com/simbot/report/wewvAijw1bobqeiw42tpvv
Dslice: https://www.raidbots.com/simbot/report/efYPTBmdvvEd6CTAn5FiBs

At first I was content with just having it send an envenom and call it a day, but after some testing, I isolated it into it's own list for clarity and added rupture and crimson tempest. While this is all fairly minor, the addition of dots into the logic doubled the gain in the cleave/hac sims. I suspect it's something timing related with the intermittent adds causing minor downtimes of some kind, but it wasn't too obvious to me on first glance.

I also checked for clipping earlier, that sucked because obviously you just generate, just a sanity check, I also checked for changing the cp required, that didn't seem to care much, forcing it on 3+ instead of 2+ was worse though, 1 was still good (basically showing that any cp you have is a-ok to dump). I also fiddled with the timings on rupture/crimson tempest for the logic but it all showed duds or indistinguishable differences, plus given how small the gain is in total, I just decided to leave it be and call it a night.

gusty ridge
#

So I was playing around with roll the bones and vanish a bit and I tried two things -
first - to not roll the bones if you have a good CtO buff, which ended up being a small gain and
second - improved the vanish->dispatch condition a bit to have the same constraint as ambush for having roll the bones (and a small improvement to have energy for dispatch before doing it)
ngl I thought the two changes would conflict with each other since one is asking to have roll the bones up and the other is dropping it on purpose, but it seems that they work well together.
From what I am seeing it results in slightly less vanish casts as well meaning it might be better to not just smack vanish on cooldown.

Anyways, I wanted to give the idea so someone could maybe improve it
https://www.raidbots.com/simbot/report/dNZHP9t5oTWQPSrpQoeGLK

Raidbots
Advanced Sim - Outlaw Venthyr - 17,257 DPS - Raidbots
half karma
#

There is something I wanted to do for some times but had no idea how to: rerolling if cto buffs> rtb buffs and ur rtb buffs arent the good ones, or rerolling if u only have cto buffs

#

look similar to what you are trying to do

gusty ridge
#

Vanish one :
actions.cds+=/vanish,if=!runeforge.mark_of_the_master_assassin&!runeforge.invigorating_shadowdust&!stealthed.all&((variable.ambush_condition&debuff.flagellation.down)&(!runeforge.deathly_shadows|buff.deathly_shadows.down&combo_points<=2)|talent.weaponmaster&variable.finish_condition) ---->

actions.cds+=/vanish,if=!runeforge.mark_of_the_master_assassin&!runeforge.invigorating_shadowdust&!stealthed.all&((variable.ambush_condition&debuff.flagellation.down)&(!runeforge.deathly_shadows|buff.deathly_shadows.down&combo_points<=2)|talent.weaponmaster&variable.finish_condition&energy>35&buff.roll_the_bones.remains>=10)

Rolling one :
actions.cds+=/roll_the_bones,if=master_assassin_remains=0&buff.dreadblades.down&(buff.roll_the_bones.remains<=3|variable.rtb_reroll) ---->

actions.cds+=/roll_the_bones,if=master_assassin_remains=0&buff.dreadblades.down&variable.rtb_reroll

gusty ridge
half karma
#

no the default one doesnt really care if ur buffs are cto or not

gusty ridge
#

it does, I am pretty sure thats what buff.roll_the_bones is, unless I have it completely wrong

#

meaning if you only have CtO buffs, that roll the bones buff will be down and it will roll anyway

half karma
#

if buff.roll_the_bones is only rtb buffs yeah, but its also not doing what I suggest

#

it would reroll if ur rtb_buffs are unders 3 seconds disregarding ctos buffs

#

what I think could be done is taking into account ctos for the rtb_reroll value, and making it reroll following the logic I said earlier

#

I m pretty certain its not the same as what you are doing

#

but what you are doing should make it reroll cto buffs less often

gusty ridge
#

it can most likely be tweaked yea, but if I understand it correctly what you are saying is somewhere in between my change and the default apl

half karma
#

kinda

knotty oriole
#

My only concern with the logic of this is that it does seem to kinda discard the knowledge that CtO buffs are preserved if they are longer duration, which would seemingly make the constraint for Vanish being &buff.roll_the_bones.remains>=10 just due to existing logic, not due to actual functionality

#

e.g. since we know CtO buffs from Vanish will be 15s duration there's technically not a reason we should care about this since they'll always be retained

#

But this is working together with your other thing to prevent it from getting "hung up" on CtO buffs

#

But I think what that means is that the RtB cast condition needs to be more aware of the CtO state and not just reusing the reroll condition

#

(Which I think is what @half karma was getting at the other day)

#

That's next on my list to integrate

half karma
knotty oriole
half karma
#

its true but I think if we fixed the reroll logic, vanish wouldnt need to be tied to rtb or cto buffs

#

what I meam is this condition wouldnt need to exist

unknown.png
knotty oriole
#

Yeah I understand what you mean, I just think it probably has to be looked at as a whole again.

#

Just need to try to avoid any conflicts between the intended changes, but I generally agree that logic should be moved elsewhere

knotty oriole
#

I am having to make a few adjustments for other talent/legendary combos for the suggested improvements to the Vanish line though so will post the adjusted line shortly

#

Some of the suggestions needed adjustment for non-WM setup (since the Flag condition being exclusive to WM is a loss but a gain when merged)

#

And merging outside of some of the legendary things

#

Looking more like something such as:

#

actions.cds+=/vanish,if=!runeforge.mark_of_the_master_assassin&!runeforge.invigorating_shadowdust&!stealthed.all&(variable.ambush_condition&!buff.flagellation_buff.up|(talent.weaponmaster|buff.flagellation_buff.up)&variable.finish_condition)&(!runeforge.deathly_shadows|buff.deathly_shadows.down&combo_points<=2)

knotty oriole
#

After some more testing, this seemed to be a gain with the latest APL logic even for QD and GS (less, 0.3% or so, but still an upgrade) for BB/GSW/Guile. Shadowdust handled by a different line and didn't seem relevant. Also had to make an exception for Deathly since it conflicts with the CP gain.

#

actions.cds+=/vanish,if=!runeforge.mark_of_the_master_assassin&!runeforge.invigorating_shadowdust&!stealthed.all&(variable.ambush_condition&(!buff.flagellation_buff.up|runeforge.deathly_shadows)|variable.finish_condition)&(!runeforge.deathly_shadows|buff.deathly_shadows.down&combo_points<=2)

finite sparrow
#

cause the dmg loss from dispatch and dps variance difference shouldnt be so extreme

lean talon
finite sparrow
#

i dont get why the dispatch sim is 2% less dps, has ~0.5% lower rtb buff uptime and a crazy high dps variance then

unknown.png
lean talon
#

is not having meld at all even a 2% loss? its probably messing up something else

finite sparrow
#

hmm yeah no meld is like 0.6% loss

knotty oriole
#

hmm

#

I think it's something wrong with auto attacks

#

Shadowmeld is triggered in the core and maybe it's not playing nice with the auto attack handling in the Rogue module

#

Can't really say for certain without looking at the actual log/debug output

#

Just the AA damage is quite different

#

in those two profiles

finite sparrow
#

if i took even 1 second to compare the differences instead of impulsively posting here i wouldve seen the difference in auto attacks

knotty oriole
#

20 fewer MH attacks

# 
  void execute() override
  {
    racial_spell_t::execute();

    player->buffs.shadowmeld->trigger();

    // Shadowmeld stops autoattacks
    if ( player->main_hand_attack && player->main_hand_attack->execute_event )
      event_t::cancel( player->main_hand_attack->execute_event );

    if ( player->off_hand_attack && player->off_hand_attack->execute_event )
      event_t::cancel( player->off_hand_attack->execute_event );
  }
#

this is in the core

#

I don't trust this

#

lol

#

We have better handling in the Rogue module for rescheduling logic but it may be conflicting

# 
void rogue_t::cancel_auto_attack()
{
  // Cancel scheduled AA events and record the swing timer to reference on restart
  if ( melee_main_hand && melee_main_hand->execute_event )
  {
    melee_main_hand->canceled = true;
    melee_main_hand->prev_scheduled_time = melee_main_hand->execute_event->occurs();
    event_t::cancel( melee_main_hand->execute_event );
  }

  if ( melee_off_hand && melee_off_hand->execute_event )
  {
    melee_off_hand->canceled = true;
    melee_off_hand->prev_scheduled_time = melee_off_hand->execute_event->occurs();
    event_t::cancel( melee_off_hand->execute_event );
  }
}
#

Still not immediately obvious why it would lose so many

#

But I think it's certainly related to this

knotty oriole
#

Still can't really explain why though, or why that APL change in particular messes it up

knotty oriole
#

@finite sparrow so did some more digging here. It's still very odd. I can't fully explain it yet but I can give some ideas.

#

Moving Shadowmeld specifically embedded into the finish list (which exits out but since it's no GCD it should come right back in)

# Finishers
# BtE to keep the Crit debuff up, if RP is up, or for Greenskins, unless the target is about to die.
actions.finish=between_the_eyes,if=target.time_to_die>3&(debuff.between_the_eyes.remains<4|runeforge.greenskins_wickers&!buff.greenskins_wickers.up|!runeforge.greenskins_wickers&buff.ruthless_precision.up)
actions.finish+=/slice_and_dice,if=buff.slice_and_dice.remains<fight_remains&refreshable
actions.finish+=/shadowmeld,if=!stealthed.all
actions.finish+=/dispatch

performs significantly better than

actions.cds+=/shadowmeld,if=!stealthed.all&variable.finish_condition
#

I think some of this can be explained inasmuch as there is ~1% lower BtE uptime for this profile so it clearly disrupts some of the normal finishing logic.

#

I still can't quite understand fully why it is so much of a loss.

#

Or why this is not much of a problem for Vanish ('embedded' Vanish is not any different in performance.)

#

I suspect it could just be timing-related as an edge case that it throws a monkey wrench in somewhere at a specifically poor time during some iterations? But I'm still digging.

knotty oriole
#

@finite sparrow Ok, I have finally figured this out. It's unrelated to Shadowmeld directly, but it was triggering a super timing specific bug.

#

@crystal crypt so does look like melding into normal finisher logic is good. But realistically just existing Vanish rules apply to it just fine. This is within margin of error for my local sim.

#

Probably will just copy the Vanish rules to keep it straightforward for all builds

knotty oriole
#

Actually going with this for now, but can test more when it's on Raidbots

actions.cds+=/shadowmeld,if=!stealthed.all&(conduit.count_the_odds&variable.finish_condition|!talent.weaponmaster.enabled&variable.ambush_condition)
finite sparrow
#

may be something entirely unrelated to apl stuff

knotty oriole
#

Essentially this was a very odd bug that was caused by Grand Melee being the first thing to apply SnD

finite sparrow
#

i also noticed there are moments where meld is being used before the cooldown is even supposed to be ready

knotty oriole
#

Which essentially never happens but with the unrestricted Meld finisher logic it could due to Vanish CtO GM into ->Meld Dispatch

#

This was the only real cause of the loss since the bug was persisted the entire fight generally due to GM

#

Took me hours to find this

#

Not pleased lol

finite sparrow
#

😄

knotty oriole
#

No idea about the cooldown thing. Shadowmeld's CD is handled in the core but I don't see anything odd about it.

finite sparrow
#

in my "ambush meld off gcd" sample, meld is cast at 6:19.334, and the next meld is at 8:19.262

#

and then it "waits" for 8:19.334 to actually use ambush

#

but it seemed unrelated to whatever is happening here

#

oh, and that this was happening to any racial, not just meld

#

very rarely (scrolling through 100 minutes of samples) lul

knotty oriole
#

Not sure

#

Could always just be a reporting issue but hard to say

finite sparrow
#

so what exactly was the issue if the first snd was caused by meld > dispatch?

knotty oriole
#

Had to do with some old Legion Azerite effect code that was stomping the normal default_value which caused it to almost always be 1.0. This had been around for a long time and that is fine. However extend_duration_or_trigger() code didn't care about that and that's newer stuff that GM and a lot of other things use. This caused the default_value to be 0.5, which resulted in the modifier for SnD being 1.25 instead of 1.5

#

This would then be extended the entire duration of the fight

#

Only way this was possible was for extend_duration_or_trigger() to be the source of the SnD buff so it was just never noticed before. GM and Premed are the only cases that can do that, and Premed is short-lived and the impact would be quite minimal.

#

And GM into finisher had not really been possible with the existing opener logic.

#

Due to the priority of SnD

#

So it required a lot of things to go wrong... which is why it only got hit in an extremely specific case like this lol

#

I just ripped out all the old code now so all should be well

#

Also implemented a way for Shadowmeld itself to tie into the Rogue module AA handling so it won't lose quite as many AAs from that either (which was a bias of like 5-6 per fight iirc)

#

But at the end of the day, yeah ~2% loss was roughly just "what if SnD was half as strong" 😛

finite sparrow
#

😄

knotty oriole
#

Yeah I didn't run into an APL that yielded that when testing today but that does seem like it could have happened for sure

finite sparrow
#

snd having a lower multiplier makes sense with the auto attack counts

knotty oriole
#

Yes, I knew it was something related to AAs but this in particular was deep into multiplier code and took me quite a lot to debug in the right spot lol

finite sparrow
#

more productive than me scrolling through samples looking for an edge case that went really wrong

knotty oriole
#

Don't worry, I did that too!

#

💤

half karma
#

I think u switched the profiles around, meaning the one using it on cd isnt actually using it on cd

#

because you modified the apl being used by the base profile

gusty ridge
#

oh did I

half karma
#

unless im mistaken i think u did that on both yeah

gusty ridge
#

oh yea, nvm then gonna delete this 😄

half karma
#

idk if its was looked at but maybe try doing it while the flag isnt being build up

#

this

gusty ridge
#

nope, still sucks besad

half karma
#

unluck

knotty oriole
gusty ridge
finite sparrow
#

was about pressing bf on cooldown during acquired axe driver

#

perhaps still worth toying with that logic cause i didnt go nuts with it

#

checked 1 dancing steel profile, saw a large loss and called it a day

half karma
#

I know he tried with bb might be ok with gsw

#

or as necro

#

bb might be the worse possible profile for it

half karma
#

doesnt look good as necro either, doesnt seem bad as venthyr gsw tho but still isnt a gain

opaque sedge
#

meld dispatch looks pog

tacit swallow
knotty oriole
# tacit swallow For outlaw, rerolling when the buff falls instead of 3 seconds before seems to b...

As a reference, the reason !up is better than "only reroll if no good buffs":

actions.cds+="/roll_the_bones,if=master_assassin_remains=0&buff.dreadblades.down&variable.rtb_reroll"
is that variable.rtb_reroll can be kept false by CtO buffs that will persist even after a reroll, so you need the !up or a duration check on the primary buff.roll_the_bones container buff to see if it has completely fallen off or not.

tacit swallow
#

Yes, that one was suggested by @gusty ridge

#

She had found it a while ago that it was a small increase to only reroll if you don't have good buffs from CTO

knotty oriole
#

I do think there could be a case where rtb_reroll would be false AND we have bad buffs AND we have good CTO buffs that are higher duration than RtB and would persist

#

But that would be a much more complex check

#

I'd suspect it could be another 0.1-0.2% or so but may not be worth

gusty ridge
finite sparrow
#

that's what #outlaw has been filled with for the last few hours 😛

tacit swallow
#

Oh sorry

knotty oriole
tacit swallow
#

That name sounds female in my native language so brain just blurped

knotty oriole
#

Is that if the cause of the odd non-refreshing bugged buffs is what I think it is

#

Rerolling in these situations could potentially be a negative

#

But idk

tacit swallow
#

Yeah, so we tried that

gusty ridge
knotty oriole
#

We still don't quite know what causes it but I do speculate it's related to duplicated rolls

tacit swallow
#

Turns out it increases the times it rolls by 0.05 every 5 minutes

#

And is essentially neutral

#

I think it's because it's very rare that a bad roll survives until 15 seconds (for vanish CTO to be relevant), and even rarer it survives until 5 seconds (for normal CTO to be relevant)

#

You would need to chain CTO buffs to get it there, and if it drops when rtb is offcooldown it'll just roll the same as if we weren't tracking that

knotty oriole
#

Yeah I think that is likely true. I think the check could be more robust potentially but I doubt it's worth the complexity of trying to find a case where it would be beneficial

#

Does look like it could be a minor gain as well to move RtB up the priority list with these new conditions since it's more timing sensitive letting it drop

tacit swallow
#

Hm, assuming the original position was good for rerolling bad buffs, would splitting the !up and reroll existing buffs conditions help?

knotty oriole
#

I think it's more just that since RtB pandemics, having it in a lower position largely didn't matter (since it didn't really make a big difference with 3s fudge factor) but when just letting it drop off, there's some minor gain to not allowing maybe 1-2s of not rerolling due to AR/Flag/etc.

#

Not really a big deal just something I noticed

#

Easy enough to just shift it up

#

But yeah may be worth testing moving rerolling around somewhere else to see if there is any difference. I suspect it'll still be neutral but I'll check after dinner

knotty oriole
#

@lean talon so @finite sparrow and I have been very bored tonight and we have done isolated RtB casts for an hour or two. For me I fully let every buff expire and raw cast every time. No CtO. No refreshes.
I did 128, Solo did 100. Based on our ranks, expected average was 41.4%
2 buff total for us combined was 39.4%
95% confidence interval for 228 samples with expected chance of 41.4% should be +/- 6.4%

So I'm leaning towards it looking functionally OK right now. Will do more rolls later. Would like to get the error down to sub-5%. But right now leaning towards thinking they either fixed any issues with it or there is something else wrong here with CtO interactions, dummy refreshes (as we discussed briefly last week,) or something else.

finite sparrow
#

For me I fully let every buff expire and raw cast every time
fwiw i also did this

lean talon
#

A little suggestion in case you werent already doing it and they havent fixed it, use the ER and SnD cancel bug to reset rtb CD quicker and with less effort

mystic jolt
opaque sedge
knotty oriole
#

In experimenting more with the priorities after the hotfix, found some notable gains for SBS for AoE sims. Putting GSW higher priority than SBS cleave lines is definitely a gain. Only about 0.1% of this is from the set bonus procs, mostly just GSW. BB gains about 0.3%.

#

Also does appear burning GSW procs with <1.5s left on the GSW buff is a small gain (0.1-0.2%) if no Opportunity proc happens.

#

@finite sparrow since it's maybe relevant to the FAQ

half karma
knotty oriole
viscid oracle
#

There's a small optimization that can be done for outlaw - the bte 5+ condition should only take into account if you would hypothetically cast BTE anyways, otherwise it'll break the 4cp finisher condition for bb build

#

Dps diff is pretty minute since BTE has to be off CD + you have to be at 4 cp with bb up

Seeing about 10 dps which is miniscule

finite sparrow
#

there is no 4cp condition for bb. it's dps neutral and not in the APL. 10 dps is within margin of error, sims will naturally vary by ±0.05%

finite sparrow
#

oh, it's in that sim too

knotty oriole
#

Could have to do with the location of it at this point

#

Also maybe specific to Necrolord

finite sparrow
#

is it not a necro+dslice only thing?

knotty oriole
#

It's possible. I don't think it should be exclusive to that, since it was clear enough it was worth doing over casting SS or letting it expire when it happened. But it was pretty rare.

#

Imagine the other specs just hit it very infrequently

#

Was a few sim/configurations where it was like 0.2%

#

Others it was neutral

#

Since it was simple enough to include figured why not

finite sparrow
#

i see

knotty oriole
#

I'd have to dig through my sims for today to find the ones where it was ahead but can look

knotty oriole
#

But yeah it's like super rare anyway

#

like 1.7 times in the sim

#

that's why I just merged it into the one line

#

But didn't really seem to make it much more complex so can't hurt I suppose

finite sparrow
#

yea, i'm assuming only necro really cares about it, but no reason to tie it to covenant.necro if its just neutral/rarely happens to the rest anyway

knotty oriole
#

0.76 times in DS

#

So yeah mostly pointless lol

#

I wouldn't have bothered if I wasn't already editing the line, but since I was

#

I figured for something like HeroRotation or whatnot it may actually be more likely in real play than a sim

#

Seems like it's possible players will miss their proc windows a little more frequently

finite sparrow
#

true, especially cause it can depend on pull structure a bit

knotty oriole
#

Also tossed in the Sepsis stuff we discussed some days ago

#

And some small ST gains for Venthyr last-second buff sniping (which was about 0.2-0.3% on ST depending on gear level)

finite sparrow
#

i also couldnt get that venthyr check to be more than neutral, though i didnt check <1s, did like <gcd*3 cause i figured 1s was too short, some iterations could be casting a cooldown during it or not have energy etc

knotty oriole
#

Was like 0.2-0.3% on my crap gear with <1.5 was marginally better with <1

#

This was 1.2s vs. 1s.

finite sparrow
#

i would also add &buff.flagellation_buff.stack<=25 etc, though i can see it still potentially being a gain at 30 stacks

knotty oriole
#

I checked and didn't make any difference

#

Seemed marginally worse for lower gear levels and neutral with BiS

finite sparrow
#

gotcha

knotty oriole
#

Running another check for that

#

Couldn't find it off hand

finite sparrow
#

thanks

knotty oriole
#

That's with a kinda meh gear set

#

Running with the default set now

#

It's really too bad the 4pc is still bugged with Flag though lol

finite sparrow
#

that it doesnt reduce the cooldown? i didn't consider that a bug before for some reason

knotty oriole
#

Doesn't deal damage either

#

Just adds stacks only

finite sparrow
#

oh, i thought it just didnt work with obedience

opaque sedge
#

ehm

#

thats bad

#

can i have the full view-extend of the bug and what is the wrongness in that

#

imma gonna go make a post on forums for that

unique zephyr
potent ferry
#

Found something interesting. There was changes recently to start doing vanish into finish, but for some reason always saving vanish for a finisher was not a gain. I tried moving the line on ambush condition that tries to protect CTO buffs: (!conduit.count_the_odds|buff.roll_the_bones.remains>=10)
To the general vanish line as well as reserve vanish for finishers with blunderbuss. Turned out to be a decent gain. Might be a gain for other legos/covs also, have not tested.
https://www.raidbots.com/simbot/report/vwiGqz4LpxGoPytKKfRdbh

unknown.png
Raidbots
Advanced Sim - T 28 Rogue Outlaw - 17,462 DPS - Raidbots
regal agate
#

nice find, side note: if you want to have the damage diff shown in the report you can usechart_show_relative_difference=1 relative_difference_from_max=1

knotty oriole
finite sparrow
#

shouldn't (!conduit.count_the_odds|buff.roll_the_bones.remains>=10) be useless now that cto buffs will remain after rolling if they have a higher duration than rtb?

potent ferry
#

Yeah I didn't isolate that as I thought it had already been tested

#

Maybe one of the recent changes made it possible now

finite sparrow
#

im also not sure why vanish > dispatch only is such a gain now for bb which seemingly conflicted with earlier tests

potent ferry
#

Or was there some other reason this change wasn't already included? Maybe it is no good i aoe

#

Maybe the !up change?

#

Possible that ambush is still worth using if you are swimming in energy also

knotty oriole
#

This doesn't seem to have any positive impact on GSW

finite sparrow
#

100% something opened it up

knotty oriole
#

Seems BB specific atm anyway

finite sparrow
#

or both of us happened to not check a bb profile

#

and the like 3 other people that contributed to vanish stuffs 😄

half karma
#

it was 100% not a gain before

potent ferry
#

My apl was bb specific, so you have to change it if you want to check other legos

knotty oriole
#

Yeah I have recreated this from scratch with some more limited conditionals

#

Not really seeing anything for other builds right now, just does seem that vanish into finisher with no Ambush is generally a gain now for BB but not others

potent ferry
#

Makes sense that it is best for BB at least

#

Nice simplification for BB then

knotty oriole
#

There's been a lot of small changes to RtB stuff, PS priority, could have been a contribution from the Vanish GM opener bug

#

Hard to say exactly what contributed

#

0.5% is notable but not a ton that can't just show up from some other dependency

#

Does seem this is pretty straightforward but I'm probably gonna double check with all the other legendaries also

#

But currently this does seem very BB-specific. Probably because it values SS way more than Ambush

finite sparrow
#

should try moving flag to a higher prio than vanish

#

currently the opener will vanish > dispatch ... build to bte and then flag > bte

potent ferry
#

Could be tied to weaponmaster/tt as well I guess

unique zephyr
#

cause it's a pretty old bug right? was happening since the beginning of the expac

half karma
#

u didnt reroll away cto rolls that were higher durations than normals rtb rolls

finite sparrow
#

if cto buffs have a higher duration than rtb buffs then you will never roll them away

knotty oriole
#

It's a gain for BB regardless of T1 talent choice

#

The CtO condition does seem largely redundant at this point, but I'll look into it more

#

Somewhat unrelated to this though since it seems to be a gain with or without it

knotty oriole
#

Seems small but probably a 0.2% gain

finite sparrow
#

looks like its no difference on st but minor gain in some aoe scenarios

knotty oriole
#

That was for ST for me

opaque sedge
knotty oriole
finite sparrow
opaque sedge
#

ideally they would want to fix stuff daily

#

but ye u ar eprolly right

finite sparrow
#

blizz add BtE to CTO in dragonflight, thanks

knotty oriole
#

BtE and SnD both get delayed a bit now which is meh but if it's a gain it's a gain I suppose lol

#

Maybe I should check if adding it to the Vanish condition makes sense

gusty ridge
potent ferry
#

Can also try to vanish ambush if bte is missing on target

knotty oriole
gusty ridge
#

what causes the sim do get 5cp from first ambush though? I've noticed it on other occasions as well

finite sparrow
#

2pc

#

check tornado trigger

gusty ridge
#

oh it doesn't show on the buffs anymore?

potent ferry
#

So bb can basically macro vanish to dispatch

finite sparrow
#

the 2pc activating is not a buff or anything

gusty ridge
#

I swear there was a buff tornado_trigger_loading or something like that

finite sparrow
#

there is, thats why i said to check

#

as it's basically your only indicator that the 2pc has activated in the sample sequence

gusty ridge
finite sparrow
#

hmmmmmm

gusty ridge
#

Nothing there, but looks like it only appears after the first full cycle

#

might be just bug in the buffs output

finite sparrow
#

oh good spot

gusty ridge
#

Also for non BB you may wanna check only ambush with broadside, its a small gain for BB (still worse than no ambush at all) but if its a gain for BB its probably a gain for other profiles

knotty oriole
#

Non-BB does seem to gain a bit more from a BtE/SnD check on the Vanish finish logic

#

Which makes sense as it doesn't care as much about using Ambush

#

Seemed more beneficial than adding a Broadside condition, but could have been indirectly doing the same thing during the opener

knotty oriole
#

Just a display issue in the sample sequence though

gusty ridge
#

I see

opaque sedge
#

im super hyped to get those numbers rolling

knotty oriole
potent ferry
#

another small improvement (0.1%) seems to be to finish at -1, stacking with other conditions, if we have tornado trigger and opportunity up, since we get at least one CP from the free BTE

half karma
#

kinda neutral tho

potent ferry
#

with my own apl for the only vanish into dispatch for BB changes though, doesn't look that that is in yet?

#

yeah, I mean it is very tiny 🙂

half karma
#

its not in yet

potent ferry
#

but I'm not sure why it is so tiny, unless overcapping on cp isn't much of a problem

half karma
#

because it doesnt happen enough

#

u rarelly use the 4p proc urself when playing bb

#

and u have to be at -1 cp for ur condition to happen

#

so its very rare

#

altough I think when i checked with celerity instead of bb it was neutral aswell

potent ferry
#

makes sense

knotty oriole
knotty oriole
#

Change is now in the nightly, so everything should be available now.

finite sparrow
#

am I correct in reading that it uses vanish > ambush in the opener and then vanish > dispatch for the rest of the encounter for all builds except deathly shadows, shadowdust and master assassin?

knotty oriole
#

MA is a little neglected for some of the recent changes I suspect (probably needs an exception in BtE line too now that we cast it more conditionally, etc.)

#

Deathly is still doing the same thing as before, just was cleaner to rip it into its own line

#

(Pretty significant loss to vanish into finisher with Deathly due to the CP gen)

#

If I have time before the 10.0 fiesta soon, I'll take a look at MA if someone else doesn't beat me to it. 😄

finite sparrow
#

how far MA has fallen, might we still be using it without the trinket nerf? lul

knotty oriole
#

Yeah it's possible

#

MA with Earthbreakers would be pretty juicy lol

#

Then again the other "problem" is that Outlaw has a hell of a lot of Crit now lol

potent ferry
#

Its vanish ambush in opener unless you roll GM I guess?

finite sparrow
#

just when snd isnt active

potent ferry
#

Now I want to sim never pressing SnD with GM up

#

But its too late and I should sleep

knotty oriole
#

tbf all this stuff is pretty much neutral and just for sample sequence cleanliness

#

Doubt we'd find any notable gains from this, the sim already only casts SnD 1.5 times per 5 minute sim on average

finite sparrow
#

pressing snd with gm active gets extended the normal way anyway

#

not like that global is wasted

knotty oriole
#

But it could just be a super minor opener update on short fights

potent ferry
#

Would probably almost only happen in opener yeah

knotty oriole
#

Would probably need some Nyalotha-level speed kills for it to happen

#

To matter anyway

#

But I guess that could happen in S4 lol

#

Actually not sure this matters since we use BtE higher priority than SnD

potent ferry
#

But vanish is higher prio than bte still right?

#

So it will vanish before first finisher, before it can proc GM