perf_prof_branch

[trim ether]

but switching back to vanilla and it works perfectly

[spiral pond]

@trim ether is it meant to explode mid air, or fall down to explode on impact?

[trim ether]

@spiral pond explode at the time of creation ie in mid air

[spiral pond]

ok. does it have a very short TTL, or create submunitions - how does it work? (only familar with the setDamage or impact approach so far)

[trim ether]

explosionTime = 0.001;```

when the main projectile hits the ground, the "helper" is created which instantaneously spawns a 2nd explosion a set height above the ground.

[analog acorn]

why not use a submunition-dispenser main projectile that can automatically poop out the explosive at the target altitude (like cluster shells but with an instant-exploder instead of bomblets)?

[trim ether]

Because this was written pre submunitions. It's worked flawlessly for years, and has only broken in a recent perf build (not necessarily the latest).

[trim ether]

Found the error/change.
The latest perf build implements fuseDistance correctly, or at least differently to vanilla. The helper rounds use the default fusedistance = 50, which they never reach before they time out. Setting fusedistance = 0 fixes the issue.

[woven loom]

Hmm for some reason going into 2D map view, at least in 3den, seems to knock the frame rates down. The more you zoom out the worse the FPS gets, and it also deteriorates with the number of units

Wasn't able to frame capture today but will the next time maybe

[worldly badge]

I thought it's very normal. You actually can see that those drawings are VERY expensive even if you uncap FPS

[woven loom]

Normal yeah but seems counterintuitive, it should be graphically less complex and not more

[worldly badge]

Like, even if I get 140 to 160 in 3D, but in main map (M key I mean) it barely touches even 50 if I zoom out in Stratis

[woven loom]

Will double check but my impression is that it's a CPU overhead/ bottleneck problem

So might have some scope for optimization! #dedmenfix

[worldly badge]

Seems the drawing forests/rocks/etc, aka drawing complex poly is very it is

[woven loom]

But it has LODs too right?

[worldly badge]

I mean in 2D map

[woven loom]

Yeah

If you zoom out, the details don't draw

But fps is still bad

[worldly badge]

Well I gotta rephrase that, if you try zoom not too out but not too in either

[woven loom]

Zoom all the way in, FPS is higher (but maybe not as high as 3D?)

[worldly badge]

You get what I mean but I don't know how to describe that using my brain after a work

[woven loom]

Yeah 😅

[vagrant zodiac]

This is normal for in game play, not just the editor, no?

[quaint flame]

It is.

[worldly badge]

That's what I mean

[quaint flame]

We had issues in SPE with that during our animated briefing

[whole cloud]

due to reasons its alot more complex

[quaint flame]

Due to bad code?

[spiral pond]

more area = more items. it still has to compute what to show and what not

[whole cloud]

2.16.151806 new PROFILING branch with PERFORMANCE binaries, v5, server and client, windows 64-bit, linux server 64-bit
- Added: Map markers are now visible in 3D in Eden Editor - https://feedback.bistudio.com/T172248
- Tweaked: getAttackTarget now shows targets earlier - https://feedback.bistudio.com/T154359
- Fixed: Unarmed units were unable to get up after being unconscious (they would be stuck in a lying on ground animation) - https://feedback.bistudio.com/T63879
- Fixed: Preprocessor would print "Invalid file name" on error, instead of printing the actual error
- Fixed: High-resolution procedural textures with too few mipmaps might not be rendered on the screen - https://feedback.bistudio.com/T180876
- Fixed: Some cases where UI to Texture would not load other textures referenced in the display on first render - https://feedback.bistudio.com/T180876
- Fixed: Crash when saving the game with a script-created #lightreflector in the world - https://feedback.bistudio.com/T179975
- Fixed: Magazines added via addXXXCargo commands were not recognized - https://feedback.bistudio.com/T74244
- Fixed: Could not scroll the player list in the role selection screen (it would always reset scroll back to top) - https://feedback.bistudio.com/T180434
- Fixed: The scrollbar in the role selection screen obscured the player mute button - https://feedback.bistudio.com/T180434
- Fixed: The player mute button in the role selection list would mute the wrong player, if the list was scrolled down - https://feedback.bistudio.com/T180434
- Fixed: Eden Editor multiplayer missions would apply the player name to the "player" unit, even if a different "playable" unit was chosen

If you don't want to use the Steam branch, the files are also available for alternative download here:
https://drive.google.com/drive/folders/15p9j7C2nHUt6NoVfChX4YFuqzFXzblJh
Note: There are separate Dll files that also need to be placed into Game folder.
📍

[unborn urchin]

• Added: Map markers are now visible in 3D in Eden Editor - https://feedback.bistudio.com/T172248
  vehicle icons are rotating too, and not correctly:

rotates with view etc

[whole cloud]

whaaaa

Yeah got it.
The code to draw icons got a rotation, but it just ignored it.
Didn't see that all the object markers were calculating and passing a rotation that was never used 😄
Thanks 🫡

[silk summit]

Since EDEN stuff is being looked at, a checkbox to spawn units as agents would be so awesome for units that are just meant to be NPCs (vendors, quest givers etc)
Could be under the "isPlayer" checkbox as "isAgent" for example, I am sure there are tickets requesting such checkbox too so if its posible please look into it

[worn socket]

Please add changeable height for the hide Module, so we can adjust its z value

[whole cloud]

Eden code stuff is looked at. Not data.
Checkbox, options, attribute changes would be data. None of such changes are planned rn

[night mulch]

Maybe you can ask 3den enhanced team for that feature

[quaint flame]

There is an agent mod for eden editor already iirc

[obsidian hatch]

Does anyone have an issue with players unfilled mags despawning when they reload? It seems that it is almost trying to auto-repack them?

[analog acorn]

#arma3_feedback_tracker message

[vagrant zodiac]

Yes - seems new since Tuesday and only affecting us on Server

[fast hornet]

also had two reports from players experiencing mags disappearing on reload (or scripted combining) today. Maybe it's a new perf v5 only problem? Have yet to test it myself.

[tribal pasture]

Been an issue for years it seems, we run stable server + stable clients, its all the same there

We suddenly started getting reports of vanishing mags recently

Could it be BE related?

Wild theory: Magazine ID precision threshold reached so game discards magazines with same IDs?

Couldn't be happening in that VR repro from the ticket though, no way ID could get that high there

[vagrant zodiac]

Don’t think it’s Battle Eye related as we didn’t have it running on the server when it suddenly started happening

Also had it with every client (not just those of us running perfect branch) so really very odd for it to suddenly start happening.

[tribal pasture]

✨ perfect branch ✨

[woven loom]

Change in a commonly used mod like CBA? Does this happen with vanilla too?

[weak panther]

Maybe the game shouldn't remove the magazine immediately after reloading starts, remove it after reloading if mag is completely empty?
i.e. not delete → reload → add back

[tribal pasture]

https://feedback.bistudio.com/T142540

Video there has in in vanilla VR empty mission back in 2019

[vagrant zodiac]

There were no updates to any of the mods we used between the last time we had ‘normal’ behaviour and the now almost 100% hit rate issue we now have.
According to steam, there was an update to perf branch but most of the clients on our server use stable. (Think the server is stable too.)
https://steamdb.info/app/107410/history/

[north girder]

someone's eating all the magazines

[heavy vortex]

Didn't seem super-plausible to me because you'd need about 6 million magazines.

Also magazines seem to retain their current ID when moved between weapon and inventory.

Not sure how the IDs are handled with locality though. Maybe you can have dupes in multiplayer.

[whole cloud]

But does anyone have a repro for the issue.
Or is it just "random in multiplayer sometimes"

[heavy vortex]

Apparently two people in vanilla VR firing two shots and swapping mags works after a few minutes.

but I'm not sure if that was DS+clients or localhost+client.

[whole cloud]

I just did it once, on first reload I got an assert on the server, trying to remove a item that wasn't where it should be

[heavy vortex]

Did that drop the mag on the client though? :P

fully expects multiple bugs

[whole cloud]

"keepInInventory" config flag.
According to comments, that should only be false for grenades.
But its false for MX magazine too.
Which causes it to trigger a networked request to remove the magazine from inventory.

Its supposed to remove the old magazine from the weapon.
But it has already been removed.

On server, the remove fails.
But, the network message about the remove, is sent out to players, before the server tries the remove itself and fails.
So even if it failed, it sent network message to everyone to delete the item from their inventory (any inventory slot)

Later the message arrives on the client, the client throws the same asserts about a incorrect deletion request.
If I ignore over a dozen assert triggers, pop, the magazine disappears on the client

[tribal pasture]

[heavy vortex]

jesus

[whole cloud]

The simple repro shown in the video.
Makes EVERY reloaded magazine disappear for me, I got a 100% repro rate 😄
Probably because debug build is a bit slower or smth

[analog acorn]

Aren't those assert triggers meant to be...not ignored?

[heavy vortex]

assert is usually a no-op on release builds.

[whole cloud]

RemoveItemCargo returns whether it succeeds or not.

And the function that calls it, is supposed to return whether the deletion succeeded.
But it just ignores the result

But the big question is gonna be, what else will implode when I fix this. Likely everything will be completely borked and its better to leave it unfixed 😄

[heavy vortex]

I wonder if other mods inherited keepInInventory or it's just the MX

[whole cloud]

Fixed that stuff, magazine still disappears on the non-server shooter :/
Plus some extra new asserts

Inventory code is a huge chaos, full of hacks and todo's from a decade ago.
Don't know if its viable to fix this. I'll probably try around and have to give up

mh

Setting keepInInventory to true for the MX magazine.
Turns my 100% repro into 0% repro, and also gets rid of all the asserts I see

Ah the comments are confusing.
keepInInventory is supposed to be true only for grenades.

Because they stay, in your inventory, while you fire them.
Unlike gun magazines which are removed from your inventory, and instead stored inside your gun's container

So this is the wrong fix for this.
But confusingly, it fixes this

Code is
if (keepInInventory || RemoveMagazine())
ReplaceMagazine();

that could be a typo, maybe its supposed to be !keepInInventory, or maybe the or is supposed to be an and.

The RemoveMagazine() line, had multiple changes to it.
Two of them were bugfixes for grenades.

So it looks like, that should only run for grenades? 🤔

But, it runs for everything BUT grenades...

I think it should be

if (keepInInventory)
RemoveMagazine(); // Remove grenades just after throwing
else
ReplaceMagazine(); // Switch in-inventory magazine with in-weapon magazine)

Running a remove as part of an or in a condition is already so cursed

[whole cloud]

😕
After a restart, that fix now seems to work flawlessly...

It definitely breaks some other stuff that seems to be unhappy about this.
But ignoring all the asserts, I don't see anything wrong.
Magazine exchanges and grenades seems to work fine now.......
yolo

huh...
If you shoot two rounds and after that instantly reload. This problematic code runs.

If you shoot two rounds, wait a few seconds, and then reload. This code doesn't run at all, so it also never gets to the bad RemoveMagazine()...
😓

[heavy vortex]

I think this might be even worse than I expected.

[whole cloud]

Mh I tried again, and now its called in every case....

[analog acorn]

Consistently Inconsistent™

[whole cloud]

I wonder if that also fixes the gesture bug, that is probably easy to repro with TFAR/ACRE animations mods?
I don't have that set up nor the time to test it. Maybe someone else can on next profiling branch :harold:

[tribal pasture]

Will have you a repro in a bit

[whole cloud]

I know a repro

I can get that mod and try it, but don't have time for that

[tribal pasture]

0 spawn {waitUntil {toLower gestureState player find "reload" >= 0}; player playAction "gestureNo"}

Execute, reload, mag gone

[whole cloud]

also in SP? or MP?

[tribal pasture]

Anywhere

[whole cloud]

The bug I fixed is MP only, so probably not then

[tribal pasture]

Yeah probably different bug

[unborn urchin]

the gesture one happens because the gesture interupts the reload

if you time it right you can even break the weapon animation/weaponstate

like the animation source will get stuck at 0.5 or whatever its at when you run it

also re: keepInInventory this thing is REALLY buggy when its used on a magazine that isn't a grenade

I had the flamethrowers in spearhead use it initially so the 'fuel' would stay in the fueltank but if you manipulated the magazine in the inventory it would behave very weird, like if you manually loaded the magazine from inventory it would behave like a regular weapon magazine and go in the weapon without being in the inventory etc

[fast hornet]

on a v5 server and v5 client i could reproduce the issue on almost every reload. But i didn't test anything else yet.

[lyric stump]

reverting server to release branch fixes reload bug

[fast hornet]

v4 server has the problem as well it seems, but not as severe as v5. Some quick tests look like v3 might be unaffected..

[kindred radish]

Funny that this suddenly appears to happen for so many people, always had this bug in Exile and even KoTH when i played (as many others).
Reported it a couple of time but never got real attention

[median belfry]

The bug existed the whole time but I guess something changed in the last few weeks and now it happens more often

[fast hornet]

never ever experienced it when i was still playing afair.. but now on an empty v5 server i can reproduce it almost 100% of the tries. Really weird.

[empty goblet]

the usual, fix one thing, reveal 3 more bugs to fix ... 🤣

[median belfry]

I'm really happy that the bug is finally getting his attention

[empty goblet]

it's always fascinating ... especially for some of the issues you knew they there, you seen them once or twice ... but were unable to actually pin them on the list

[hollow canyon]

Sounds like my code, lol

[median belfry]

The magic of Arma

[north girder]

people are getting arma'd even in coding

[hexed verge]

is it part of the current perf_prof build if the game crashes it keeps running and everything is moving, but you are locked to a dialog box if you want to create a crash dump or not?

that caught me by surprise

[patent dome]

Do you have the debug parameter set? That has a dialog box that asks if you want to create a full crash dump or a regular minimal crash dump.

[hexed verge]

yes, i run arma always with full debug and showing script error flags.

it was not the dialog box that surprised me, but rather the game was still running and i could see people walk around and even talk to me over ACRE/Teamspeak

[heavy vortex]

I would guess that's a minor-thread crash.

[whole cloud]

Guess it could be a crash in a thread or so.. 🤔
Send me a dump and I'll find out why 😄

I don't know if the crashdump creation does a kill at the end. I have never seen that happen

[hexed verge]

i knew i should have pressed yes instead of no..

actually got the mdmp and bidmp

via pm?

[whole cloud]

anything you like

Though snailmail might take a bit too long

[hexed verge]

i shall fax it

[quaint flame]

The true German way.

[whole cloud]

Indeed crash in a thread, in XAudio thread that we don't control

[whole cloud]

2.16.151824 new PROFILING branch with PERFORMANCE binaries, v6, server and client, windows 64-bit, linux server 64-bit
- Added: Eden Editor markers can now be placed in the 3D (non-map) view
- Tweaked: Eden Editor object icons now rotate with the object
- Tweaked: Eden Editor map markers now become visible when selected even if their alpha is 0%
- Fixed: Eden Editor marker placement preview icons did not exactly match the markers (slight size difference and missing shadow)
- Fixed: Submunition shots could duplicate their effects (particles, impact craters) in multiplayer - https://feedback.bistudio.com/T177010
- Fixed: Partially filled magazines could disappear after reloading in multiplayer - https://feedback.bistudio.com/T142540

If you don't want to use the Steam branch, the files are also available for alternative download here:
https://drive.google.com/drive/folders/15p9j7C2nHUt6NoVfChX4YFuqzFXzblJh
Note: There are separate Dll files that also need to be placed into Game folder.
🗺️

[hybrid wasp]

Games crashing, 4 times back to back, just got on about 10-15 minutes ago. Kept crashing on the configure keybinds setting menu, and also crashed for seemingly no reason as soon as I got on an MP server

Is not crashing on main branch

[whole cloud]

Can you send me one of the crash reports?

Configure keybinds is this?
Fine for me

[dense steppe]

I was tweaking some settings on my server and just had the server hang with the newest perf binary. it never fully crashed but it did red chain the users who were on, and completely locked up until i shut it off

Unfortunately, the log doesn't show anything, it just looks like a normal shutdown

[hybrid cradle]

cba settings or server settings

[dense steppe]

CBA settings

specifically ACE

[hybrid cradle]

sounds like it could be a gui problem then

tho that wouldnt explain a server hang

[whole cloud]

I also didn't touch any gui like that, so that'd be weird

[hybrid cradle]

just two things in menus

i can shit around with perf later

[dense steppe]

I personally don't think its related, as it only happened after I started getting more users on the server

[whole cloud]

The time to revert the build already ran out. So we're stuck with it till tomorrow

It'll crash when someone shoots a projectile without a shooter. Didn't expect that to be possible 🤔
But you can probably scripted create them without one

[hybrid cradle]

you can

i rely pretty heavily on it

[fast hornet]

hm yeah.. also had my v6 client just crash.. no error or anything, just exited.
The v6 server is still running though and i can reconnect without issues.

[whole cloud]

yeah just mineCreate command would do it.
Servers will also crash when a client does it

[hybrid cradle]

wait that's a command

[analog acorn]

mineCreate isn't, but createMine is

[fair swan]

Im crashing too. Playing warlords and anytime an object is purchased or if another players parachute lands nearby game froze. Works fine without profiling

[random isle]

yep used by mineSite module

[fast hornet]

i can run e.g. _mine = createMine ["APERSMine", ASLToAGL getPosASL player, [], 3]; just fine without it crashing.. v6 client and v6 server.

[whole cloud]

We reverted steam, google drive still has the broken one. Fix it tomorrow

Still looking for some crashdumps.
To confirm that what I think it is, really was it

[woven loom]

It's like watching an episode of House but for Arma 😄

[whole cloud]

v6 is back now

[fast hornet]

btw. would the reloading/magazine despawning fix need the client to have v6 or only the server? Or both?

[whole cloud]

Serverside only could be enough.
I have only tested with both

[fast hornet]

Thanks

[marsh gyro]

Did perf update to 2.16.151824 - ?
My server is showing this version but this seems a few numbers higher than what the last update shows.

[wet turtle]

is there a reason why
diag_captureFrame, diag_captureSlowFrame,diag_logSlowFrame
are disabled in current dev branch?
@frigid yarrow @final mantle

[restive pilot]

Probably due to the hotfix?

[whole cloud]

[whole cloud]

It sounds like the magazine disappering fix has worked so far?
QA also confirmed its good, and so far I have not heard of any negative effects?

[weak panther]

Server on prof, clients with and without Prof cannot pero bug more

[ashen helm]

"Eden Editor object icons now rotate with the object" why I thought it is a bug? Anyone requested this? What's the point of this?

marker icons are rotated in 3D view but on map are still in default position, this is hella confusing

[quaint flame]

Map is always oriented north. Why rotate?

This is a good thing. Imho

[weak panther]

There is other mags disapp problem
the magazines disappear when reloading starts, unfull mag is back on complete
but if at this moment abort reloading with a gesture, the magazine will be completely lost

[weak panther]

Can be done some protect for this?
This can be reproed using mods: tfar animations, and some malee mod

[whole cloud]

This feature was already present in Eden code. But one of the functions didn't pass the parameter along.
As I needed that parameter for markers, I thought I'd just keep it.
Why, whats the problem with it?

but on map are still in default position
What do you mean?
Markers are rotated the same on map and in 3D

[tribal pasture]

Seen positive feedback, but can't guarantee 100% yet

Prof server, stable clients

[unborn urchin]

these mods can fix it by checking gestureState before overiding

[fast hornet]

had no player reports after i changed the server to v6.. so might be fixed.

[empty goblet]

way more crashfixes wip/done, sadly no hotfix/profiling today, crossing fingers for next day

@grim also that question is for #dev_rc_branch aka channelception ;)

[woven loom]

Frame cap, it's the HUD/map drawing by the looks of it https://imgur.com/9W7jhpm -- this is when the FPS was around 40 constantly

With some vanilla maps, going from being zoomed all the way out to in sees the FPS go from 70-ish to 40-ish at medium zoom and then back to 70-ish when you zoom all the way in

[quaint flame]

That's known.

[random isle]

why is the map drawing instead of being pre-cached from start ?

[wet turtle]

@empty goblet i couldnt find, #ToManyChannels!

[empty goblet]

@grim i could add more ;) like #complain_about_VON:)

[quaint flame]

Most likely because of different scale.

[wet turtle]

#complain_about_toManyChannels? :)

[whole cloud]

lol who cares about map

[shrewd moss]

So, let's suppose an average player looks at the map 10 sec every 10 minutes of play.
Arma 3 has an average of 9708 players.
If we do 10:600=x:(9708 * 24*60 *60) we get that everyday the map is looked for about 14000 secondw
Let's suppose that the average cpu has and tdp of 65W.
That means that the map is using 9100W or 91kW per day.
If you could optimize the map usage by 10%, that would result in 9 kw per day saved worldwide, or about 3300 kw per year.

Let's suppose a carbon footprint of 0.207 kg of Co2 per kw.
An optimization of 10% of Arma 3 maps could save 683 kg Co2 worldwide each year.
Let's suppose we have to wait at least 3 year for Arma 4. This optimization could save at least 2000kg of Co2 worldwide.

The environment is in your hand dedmen lol

It's 11pm, I know nothing about electronics, I probably made a tons of mistakes and I should be sleeping because tomorrow I'm working....good night

[fickle geyser]

You're not taking into account that if he optimizes the map people will just pull more FPS. Effectively using same amount of energy.

[heavy vortex]

Now adding option for servers to pause when there are no players would probably help more.

[fickle geyser]

Yeah, fix LODs, more important.

[random isle]

What about mini map ?

[woven loom]

not noticed any effect

[woven loom]

On the contrary, if it caps GPU well below its capability then that's a lot of power being saved
Feature, not bug! 😄

[woven loom]

what is

[rain moth]

but that's the default behaviour? it stops the current mission when last player leaves?

[heavy vortex]

I mean actually pause, not stop the mission.

[silk pewter]

wouldn't dynsim do that? but I see what you mean

[hybrid cradle]

time variable still increments which i believe is the main desire here(?)

[silk pewter]

ye, can't fight that through script

[hybrid cradle]

time = time-1 :)

[whole cloud]

DayZ doesn't pause, but it has option to reduce FPS to like 10.
Does sound very smart indeed. Someone should see if there is already a FT ticket for that so I can put it on some list

[empty goblet]

lot of games have options to define FPS in UI (menus/editors) separate from ingame 3D FPS, rarely i seen even extra option to define FPS while using fullscreen elements like 2D map ...

while FPS overall throttling is doable via application like MSI Afterburner ... being able define specific aspect of game to spent less power on is nice for mobile users ...

i seen surge of this thanks to STEAMdeck and similar handheld gaming devices released

similar goes with pause (for singleplayer games) until the mobile gaming became thing, it didn't often 'save' enough resources now developers pay more attention to do near full stop

[quaint flame]

maxFPSIn3DEN?

[empty goblet]

ye but then probably to cover all cases like maxFPSin2D/3D/UImenu/3den etc.

[light cargo]

valve source engine always had a power saving measure of lowering the fps when unfocused, options for that were added in 2013 to the hl2/tf2 branch

it feels too late in the dev cycle to add this feature to arma :/

[heavy vortex]

The A3 client does it already, at least in fullscreen.

[autumn timber]

But is it that much of an issue, realistically? Maybe that's just me, but when I'm playing, I'm... playing. If I want to put Arma in the background for 30 minutes, I'm just going to turn the game off, and do that other thing that i wanted to do, instead. I don't think alt-tabing for 30 seconds impacts my electricity bill that much.

[fickle geyser]

borderless/window too

[fickle geyser]

more important for servers with persistent missions

you don't need these 100 frames when there's 0 players imo

[autumn timber]

Yeah, I agree on that one

[fickle geyser]

I'm actually happy that there's noPause that disables that as otherwise it would freeze the stream when you alt-tab for example

same for noPauseAudio or whatever was the param

[autumn timber]

Saved my butt a few times 😉 (when playing persistent pvp missions)

[heavy vortex]

It'd be an issue if it didn't. I often forget I have Arma open while testing stuff.

[woven loom]

I think a lot of people have the game running in the background while scripting in notepad/vscode etc, then alt tabbing to test, etc.

I think it's a common joke how half of the thousands hours in Arma are spent looking at notepad++ 😛

[hybrid cradle]

using n++ is why you're spending half your time not looking at arma

[woven loom]

I didn't know vscode and n++ were the same thing

Anyway, bizarre tangents aside, I think it's a very common situation 🤷‍♀️

[light cargo]

they're not but they're used as code editors

[fickle geyser]

ADT

[heavy vortex]

ADT is great but typing code into it is not the best experience :P

[woven loom]

I'm aware, I was being sarcastic

[restive pilot]

It's still vanilla textbox

[whole cloud]

That's the issue! You need to mod in a better textbox 😄

[hybrid cradle]

love adt except when i accidentally press ctrl+w

[restive pilot]

How do you "accidentally" press ctrl+w?!

[silk pewter]

accidentally.

[hybrid cradle]

keybinds before menu closes at low fps

it happens like once a month

tres annoying

[empty goblet]

lately i noticed some games already have settings for maxFPS in focused mode vs unfocused mode vs minimized mode

[trim ether]

Looks as though the missing mag bug is still in the Perf build. Our server runs Perf build and we had 3 instances of players loosing mags last night. All the players that lost mags were running Perf build clients. 😦

[whole cloud]

have a repro?

[trim ether]

Yeah, but unfortunately its large. I'll DM you.

[magic belfry]

@trim ether Is this a bug in which players reload and the mag appears briefly in their inventory then vanishes?

We're having that happen with full mags but not all the time?

[trim ether]

@magic belfry We saw it when reloading a partially full mag. It was flagged about an hour into a mission with 16 players. I tried to reproduce for Dedmen using a single player on the same mission/server, but failed to generate the error.

[whole cloud]

It is very likely that there are more bugs that cause the same symptoms.
There is also the reload during gesture but that's also not fixed.
The other cases will probably not be investigated

[magic belfry]

Im just updating the server real quick and ill see

I suspect it was happening for us due to the outdated server now.

[empty goblet]

1.54.133701 profiling & performance v3 server and client, windows, (linux later)

• crash fixes, issues fixes, some tweaks and optimizing (for client too)
  https://www.dropbox.com/sh/582opsto4mmr8d8/3BSy9PdRGm

[wet turtle]

@empty goblet, How does one start profiling binary with BE? :P

[empty goblet]

stupid q, backup arma3.exe or arma3server.exe and replace with the respective one you download

@wet turtle https://www.dropbox.com/home/Arma3?preview=how_to_readme.txt

[wet turtle]

Well im asking as A3Launcher can launch a3 profiling without doing so it just passes an argument to arma3profiling.exe.

[empty goblet]

no idea, if you think it's bugged, hit the launcher devs with bug report ;)

definitely thinks my order menu said seamless switch support :)

[wet turtle]

Well launcher works, but i dont wanna use it :P batch files are faster :P

[empty goblet]

please report me any crashes, also make sure your servers/mods using latest callExtension versions and not 16 versions old extDB2.dll for example (giggle)

[whole cloud]

2.16.151860 new PROFILING branch with PERFORMANCE binaries, v7, server and client, windows 64-bit, linux server 64-bit

- Added: Vehicle marker lights now animate with animated memory points
- Fixed: Ban reason was not displayed on the client being banned (but kick and BattlEye kick reasons were) - https://feedback.bistudio.com/T181310
- Fixed: Server kick/ban commands now support whitespace in usernames by wrapping them in quotes - https://feedback.bistudio.com/T181310
- Fixed: Performance degradation when many scheduled scripts are active
- Fixed: Missile targeting sensors did not check range/angle limits (anti-radiation missiles could target radar emitters that were behind the missile, out of range and whose radar did not even reach the missile) - https://feedback.bistudio.com/T181124
- Fixed: removeMagazine(Global) on a grenade would make the grenade 'type' unusable, even if more grenades were available - https://feedback.bistudio.com/T79813
- Fixed: Removing all throwables via addMagazineCargo/addMagazineAmmoCargo would leave a ghost grenade behind - https://feedback.bistudio.com/T181275

If you don't want to use the Steam branch, the files are also available for alternative download here:
https://drive.google.com/drive/folders/15p9j7C2nHUt6NoVfChX4YFuqzFXzblJh
Note: There are separate Dll files that also need to be placed into Game folder.
🤧

[hybrid cradle]

any more specifics about the scheduled scripts fix?

[quaint flame]

Something about sorting the queue etc. Can't find dedmens post about it.

[analog acorn]

#arma3_scripting message

[hybrid cradle]

zamn

[upbeat thistle]

There is a problem detected aftet the update. We cant see the players list on infistar menu if we using the profiling version. If we use the normal version then the list works fine

[whole cloud]

I don't know how the infistar menu works. Can you maybe get an infistar dev to tell me what reasons might be?

The only thing related to players list that I see would be the kick/ban commands?

[upbeat thistle]

yeah already opened a ticket on infistar im waiting for their response and ill be back with a feedback

i hope at least..

[inland dew]

should this help with server FPS (over time)?

PvE

[whole cloud]

Only if you run bad scripts that fill up the scheduler

[knotty wraith]

am I the only one who lost the script console if I log in as an admin?

I'm putting back arma3client_x64_performance_216_151824_v06 - the script console is working

[whole cloud]

Are you running it on server too? or only on client?

[knotty wraith]

I also run it on the server

[whole cloud]

so v7 on server, and v6 on client works.
v7 on server and v7 on client doesn't work?

[knotty wraith]

yes

[quaint flame]

Haven't you tested it with like 20k scripts? Guess that's not too relevant.

[whole cloud]

yea

[fickle geyser]

good, I don't need to think about releasing improved liberation anytime soon

[kindred radish]

Just tested newest client prof with 2.16.0.151824 prof server and its broken with ExileMod.
Player doesn't seem to get initialized properly, no errors in server rpt or client rpt.
Reverting client back to stable solves the problem.
Gonna test both on newest prof version if there is the same error.

Edit: Same error with newest prof on the server.
You can move the camera, still hear your own steps, but it's like you are a "cameraman midget".
https://cdn.discordapp.com/attachments/767092419799547954/1247560632056807474/image.png?ex=666078e1&is=665f2761&hm=a8c658d351c172974ef30832cf7641863921c9647ed6c9094aa854d2abe9a639&

[knotty wraith]

class CfgMusic {
    class Podnimaya_znamya {
          name = "Podnimaya_znamya";
          sound[] = {Podnimaya_znamya.ogg,db,1};
    };
};

playMusic "Podnimaya_znamya" - v7 on client doesn't work
v6 on client works

[stray prawn]

A3Wasteland is being stuck at player initilization with the newest v7 Server files, with the v6 server files initilization + client v7 it was working. cant find any errors in the server rpt or in the client rpt either.

[inland dew]

can confirm - script console gone when logged in on the server, both running v7...

[whole cloud]

sus

We clearly are doing too many things if prof keeps breaking.
need to do less stuff

[hybrid cradle]

couldnt some of this stuff get put on dev branch? suppose less would get caught tho given theres no server branch for dev

[rose moth]

Prof v7 reverted to v6 for now.

[fast hornet]

Just ran some tests with ExileMod:

• Client

• stable - works

• v6 - works

• v7 - broken

• Server:

• stable - works

• v6 - works

• v7 - works

ah, @rose moth was faster 😄

[rose moth]

Dev-Branch is less effective these days due to its population, and especially it's not really a multiplayer-active branch, limiting what can be tested there.
Prof has become a tad more experimental in recent years, but it also brought many powerful things. So it's a small price we pay for progress at this stage in the game's life. We could add an actual QA smoke test to Prof, but I fear it would slow down progress quite a bit. Having said that, we'll try to break this current streak 😇
Thank you all for the swift testing and your patience when these interruptions happen.

[hybrid cradle]

i personally dont mind it at all given it gets reverted quick enough to not interfere with anything major tbh

[fast hornet]

i mean.. it's not ideal imho to push these highly experimental builds onto players. Many of them are using them for the possible performance gains.

But then again it's usually just a few tickets.. a look in here, ah.. new build. And telling the players to revert to stable.
All in all it's better compared to stagnation i guess.

[hybrid cradle]

small price to pay for making the scheduler slightly less shit

[rose moth]

btw Dev: 0.02% of players, Prof: 0.83% (client)

[hybrid cradle]

i dont touch dev branch bc of no mp stuff atm

[rose moth]

highly experimental builds onto players
Fair, but they opted into an experimental branch. I know the waters have kind of muddied, since a lot of people just get pushed there "for moar framez" and servers often run with it.

[fast hornet]

these recent perf builds from v3 onwards had massive gains in terms of server script performance for ExileMod as far as i could tell, so i'm fine with that anyway 😄

v6 has been running great on our live server for two weeks now

[knotty wraith]

don't touch serverv7 - everything is basically up and running for me)

[fast hornet]

the (v7) server logs were strange though.. more pointing towards fucked battleye/steam session ids and stuff.

[autumn timber]

Prof: 0.83% (client)
Woah... I wouldn't think that almost 1% of the userbase even knew about the perf branch yet alone to actively use it!

it's not ideal imho to push these highly experimental builds onto players.
IMO if someone was smart enough to use a nonstandard branch, they will have no issues with reverting to the regular branch, in case of issues, tbh

[kindred radish]

Maybe a simple test setup to test such builds before with some popular mods/game modes would help.
Just a basic test like "can i join the game and do the basics" 5 minutes and done since most of the issues were visible quickly.

[fast hornet]

Or maybe releasing the new builds only to the google drive and announcing it in here (with an actual @ ping) and then waiting something like 12-24 hours before pushing it to steam could be nice.
Slapping a new build onto a dev server and test client only takes a few minutes

And/Or if you could provide a static link that always contains the current perf arma3server_x64.exe/arma3_x64.exe .. that would make it really trivial to auto-update a dev server for example

[whole cloud]

Just a basic test like "can i join the game and do the basics" 5 minutes and done since most of the issues were visible quickly.
Issues like that happen very rarely though. More often in the last weeks, but overall its very rare. Most errors are more obscure and harder to run into

For example this build that's clearly oofed.
Tried to repro the issues, launched a server, launched a client, joined server and started a zeus mission.
Everything went smoothly, no issues whatsoever.
Building the server and client builds for it made this take about 20 minutes. And it was basically useless.

[analog acorn]

I have noticed a few high-profile people in the community / devs recommending using prof as standard, either for actual performance or to use new features, to people for whom it probably wasn't the best idea.
While it's fine most of the time, I think this kind of shows that the messaging might need to be a bit clearer in such cases that this is experimental and not always "just stable but better, so you should move your entire server to it".

[whole cloud]

Cannot reproduce..
v7 dedicated server (running locally), and v7 client. I can join, run a zeus mission and use debug console.

Debug console is fine, server exec works too.
My camera is not stuck in the ground
I don't have infistar problems (also don't have infistar though)
A3Wasteland....

[whole cloud]

Doesn't work means it doesn't play audio, or it prints some error to RPT?

[knotty wraith]

I looked - everything is correct
client v7 - error on RPT

File: E:\arma3files\Users\Dimon%20UA\mpmissions\evo.yulakia\podnimaya_znamya.ogg.wss not found !!!

but I repeat - before v7 it always worked

[whole cloud]

A3WastelandI can reproduce! That's something atleast

18:17:53 Scripting function 'bis_fnc_execvm' is not allowed to be remotely executed
18:17:53 User dedmen [BIS] (76561198049800) tried to remoteExec a disabled function: 'bis_fnc_execvm'
Error in server log during start.
But A3Wasteland uses CfgRemoteExec and which doesn't whitelist that function, and specifically mentions that it should never be whitelisted.

Why is it used if its not allowed? 😕
This looks like a obvious thing that could cause issues like this, but seems its just normal

Player is waiting for its playerdata, that A3W_fnc_requestPlayerData from server should send over.
The execFSM "call.fsm" is executed, but the code passed as argument to it, is never executed.
What even is that mess.
Why does it have
[x, y] execFSM "call.fsm"

when the fsm only contains
x call y
why run a fsm that just calls a script function 🤯

Maybe I broke execFSM, and maybe infistar and exile are doing similar weird stuff

Added call.fsm by KK to execute non-scheduled code from scheduler.
If you want to run unscheduled code, you can just wrap it in a isNil {}, instead of this mess 🤯

Yeah the FSM's are not executing

Hey I found a copy-paste mistake in the FSM processing from 2019

Found it.
The pain for compilers to always prefer int as default datatype.
Even when its an enum that stores the possible values 0, 1, 2
"Oh 2 you meant? Sorry! I thought you meant -2"
"Oh you want me to compare your -2 and see if it matches the last enum value (2), well it clearly doesn't"

Finding this stuff with QA testing would be hard. Today was the first time I saw someone use execFSM as an obscure way to do a isNil
QA might've only found that by playing a campaign mission with scripted FSM as mission flow

[fickle geyser]

Whole East Wind is using FSMs IIRC.

Or Tac Ops missions.

[whole cloud]

jup. got it.
That's a bigger one in many places. Thank you

[kindred radish]

Exile also uses something similar to Wasteland to request the player (the fsm also handles loading screen etc).
https://github.com/Andrew-S90/ExileMod-Files/blob/main/Client/exile_client/fsm/login.fsm
https://cdn.discordapp.com/attachments/646531827397296138/1247598345439150170/image.png?ex=66609c00&is=665f4a80&hm=c5aa11d76f0ee7d7ccbf89175f38dad6589d82378713bd6309e642fbcd30d09b&

[whole cloud]

Fun legacy code.
How to check if a string is empty.

1. string.IsEmpty()
2. string.size() == 0
3. !string.data()

2 of them are sensible, and 1 is just sugar for 2.
But old code likes to be funny.

[inland dew]

several server owners, with completely different missions, v7 client = no debug console after login

can't tell you more than that

[heavy vortex]

Currently wasting a few brain cells trying to figure out whether 3 is legal.

[knotty wraith]

add the ability to use scripting commands from a future patch 🤪

[heavy vortex]

I feel like prof percentage with active players must be higher, because it's a much bigger proportion of the client RPTs I see.

but maybe that's causal :P

[tribal pasture]

2013 tech

Achieving stackable per frame non-scheduled execution took years

I think people thought FSM are always unscheduled

delayless_loop.fsm was a thing before onEachFrame and then finally KK's brakethrough well into A3 with perfect OEF using createVehicle'd triggers

[merry shuttle]

It isn't A3Wasteland which causes the remoteExec disabled function 'bis_fnc_execvm' error but some BI script which calls initPlayerServer.sqf https://community.bistudio.com/wiki/Event_Scripts#initPlayerServer.sqf

[empty goblet]

well ... the 'performance' marketing helped bit with the awareness 😁

[whole cloud]

2.16.151868 new PROFILING branch with PERFORMANCE binaries, v7, server and client, windows 64-bit, linux server 64-bit

The music thing and FSM thing are fixed.
Debug consoles I still don't know what's up, its possible that the others somehow influenced that one and it might also be fixed?

[light cargo]

hmmm, maybe creating some engagement with multiplayer parties could help at all towards some dev activity?

like i dont know, stress testing you cant really do in the office or with simulations

or perhaps in the event everyone is out of office and working from home, could have a volunteer qa session with a big question mark on sweepstakes

idk, something that might make it worthwhile

i would devtest but i dont play a3 often and i dont have storage for a p drive dev install, and retail mp is where is it at

[restive pilot]

Maybe you broke the scheduler?! 😅

[light cargo]

yet i have storage for 4 gta iv/eflc installations

and both retail ms flight simulator x and steam fsx (without addons)

anyways, sorry for the derail

[rose moth]

The thing with Dev is that it's a full branch, including data, which typically makes it MP incompatible with the masses.
But Dev is still useful for future update data testing, and for modders to preview how the future update may affect them.

Prof is a no-data branch, which is part of what makes it MP compatible with main branch.
And then there's still RC branch before major updates as the true test.

Your ideas are good, but it's more about what's still feasible and reasonable for a 11 year old game 🙂

[light cargo]

yeah i understand that

perhaps dev could be data compatible for a small period for some tests and working in code especially after a release?

would that help a bit with testing?

having a small period of no data changes after a release

[restive pilot]

You can test dev with other dev builds
Tho there is no dedicated dev build afaik

[light cargo]

is it too much to commit to?

right fair enough, i guess it shouldnt be too much of a hassle to switch steam branches then

[light cargo]

isnt arma3server included in the game anyways? so basically dev builds do have a dedicated server, if that's what you mean?

arma3x64.exe can do dedicated server functions via -server

[light cargo]

gotcha yorisyan, its understandable.
right i forgot about rc but those are like, ephemeral in a nutshell.
i guess profiling is the place where "how far can you go without introducing compatibility breaking changes"
there's not much going into dev lately because of the mantainence status of the game, how many years of low traffic now?
(sorry for another ping >_<)

[rose moth]

I haven't check detailed traffic over the years, but it just became less relevant over time. During primetime development there were daily updates, DLC pre-release testing, etc. so it was very enticing to try. And the flat % number also doesn't really account well for people who occasionally check into the branch. I don't really see it as a problem btw 🙂

[whole cloud]

but it was only FSM's.
Debug console doesn't use FSM (afaik??) and i couldn't reproduce it being broken before fixing the scheduler

[light cargo]

quick history-related question that you might know:
why the shipped dlc formula with upselling has been dropped? was that the plan all along for the game lifecycle? it seems that the dlc bundles still exist but got delisted

[rose moth]

I'm not quite sure what you mean, but let's stick to Prof in this channel. Ping me somewhere else on this one 😉

[restive pilot]

Well I mean you said you optimized the scheduler to only sort a portion of the scripts right? (only 256 per frame iirc)
What if it only runs that 256 and not the rest? (Or conversely, runs those 256 scripts once and not anymore)
Also another thing about script sorting. I assume script age means when the next script should run? (e.g. sleep 1 means "age" is 1)
But if you only sort a chunk that means a script may never run at the correct spot? (Unless you define the age of the whole chunk which I assume you did)

[knotty wraith]

description.ext

//enableDebugConsole                = 1;
enableDebugConsole[] = { "76561198XXXXXXXXX", "76561198YYYYYYYYY"};
```this is how the script console works

[whole cloud]

What if it only runs that 256 and not the rest? (Or conversely, runs those 256 scripts once and not anymore)
It would get the next 256 chunk next frame, because after executing all the executed ones age's are zero, and they'll be sorted at the end next round.

I assume script age means when the next script should run?
No age is time since it last ran

[whole cloud]

You mean steamid whitelist is not working?

[knotty wraith]

enableDebugConsole[] = { "76561198XXXXXXXXX", "76561198YYYYYYYYY"}; - work
enableDebugConsole = 1 - not working

[whole cloud]

I set it up using mission.sqm setting to enable debug console only for logged in admins (the = 1). It doesn't show up. ✅

The check for that is
call (missionNamespace getVariable "BIS_fnc_admin") isEqualTo 2 // logged in admin
Or on server side
admin remoteExecutedOwner isEqualTo 2 // logged in admin

call (missionNamespace getVariable "BIS_fnc_admin") -> 0
serverCommandAvailable "#logout" -> false. Tadaa.

Good old off by one, or alternatively, using the wrong variable name.
That's what I get for refactoring old code only half way

{
  if (!cmd.BeginsWith("#"))
    return false;

  cmd = cmd.Substr(1 /* skip start # */, cmd.FindFirstOf(" ") /* till first space, in case it has arguments */);
  auto type = FindNetworkCommandType(cmd);
}

Such a tiny thing just snowballing into a bigger thing

Fix for it will be pushed Friday morning. Essentially "serverCommandAvailable" always returns false.

[vivid rune]

Is this maybe also an exploit to gain admin rights by kicking somehow the admin/zeus and then become admin rights?

[whole cloud]

Bug reports that are not profiling branch related, please to #arma3_feedback_tracker
Keep this channel about prof branch related stuff

[light cargo]

RIP

[light cargo]

ohh, thanks for info

[fair ore]

Noticed I couldn't lock air targets with a manpad on v7, haven't had a chance to repro with no mods so not gonna feedback tracker it just yet. Wondering if anyone else had that.

[knotty wraith]

my player also complained yesterday that CUP_launch_Igla is not hitting the target
I'll try to check it myself today

[fair ore]

Since it's reverted, I guess we can't test?

[whole cloud]

its not reverted

Missile targetting was broken previously. It could target things out of range/vision angles. Its possible that some mods are missconfigured with bad ranges, and just never noticed

[heavy vortex]

uh oh

[knotty wraith]

manpads do not work, they lock the target only if it is low above the horizon and far away (CUP)

[fair ore]

I'd trust CUP config but also I tried the titan launcher and couldn't lock

Just didn't do it with no mods loaded

[full nova]

is the air target hot / flying?

[fair ore]

It was! I'll test properly when I get home with all vanilla things

I'm pretty sure I wasn't loading any mods that change the titan AA launcher but I want to be sure

[full nova]

because i remember some modded aa launchers if you too close to the target they will not lock (RHS + Ace), not sure about now

[fair ore]

Yeah these were both close and far since I was uh

Being strafed repeatedly and desperately trying not to die

I should clarify

It wasn't missile guidance, it was being unable to even lock it up with T targeting

But I won't call it a real bug until I can repro at home with vanilla

[whole cloud]

But also the fix I made is after lock. When the missile is already flying. Or atleast its supposed to only be then

[fair ore]

yeah

[neat sorrel]

It seems the last patch breaks CBA, ACRE and maybe more mods. Tested on a Linux server with main and profiling branch clients and happens on both cases.

[whole cloud]

Explain what "breaks" means

[neat sorrel]

oh sorry, you're right, I didn't give much explanation.

Since the new patch ACRE2 does not work on any of the clients. The TS client automatically switches to the "ACRE" channel, so something is working, but the radios are not. I'm not really sure how to provide more information about the problem, I don't see anything related to it in the logs, but I'd love to help solve it by giving you any necessary information

[whole cloud]

What does it mean when the radios are not working?
And what do you mean about "breaks CBA" ?

Can anyone else confirm that ACRE is broken?

[neat sorrel]

When you press the PTT button, the radio doesn't work. No sound, no text box on the right corner of the screen, no simulated interference, etc.

[neat sorrel]

Sorry, about CBA can't say much. I've read on #server_admins another person has been having issues since yesterday with CBA on the profiling branch, and I thought it would be related, but maybe this is my mistake.

@craggy kraken can you provide more info?

[whole cloud]

I am, however unable to view the debug console in the ESC menu or to alter the server-side addon settings
Well yes.
#perf_prof_branch message

[neat sorrel]

maybe this fixes it, right?

[whole cloud]

I don't see how server commands checking has anything to do with ACRE radio's working. it shouldn't

[craggy kraken]

Yup, it was definately ACRE that failed. Only on the server side tho. Client side it works as a charm.

[neat sorrel]

me neither, but i'm really lost on what could be the problem

[craggy kraken]

I've no idea why.

My suspicion is that it has to be related to the profiling branch, as it was the only thing that changed.

[neat sorrel]

Yes, that's true. If I'm not mistaken, since saturday the only major update was on the profiling branch, not on any of the mods

[craggy kraken]

Yeah, I can confirm that no mod has been updated since at least the 30th of May.

There was an update of Steamworks Common Redistributables. But I'm not aware that it could affect how ACRE works server-side.

The mission used, was one we had played multiple times before. So I don't think it's the culprit.

[empty goblet]

any feedback on v3 yet ?

[whole cloud]

So ACRE is only broken if profiling branch is installed on server side?
That atleast makes it not too critical, but I got nothing to go on about why that would break radios.

It wouldn't be the scheduler changes because ACRE is a unscheduled mod.
The serverCommandAvailable is currently still broken, but why would ACRE depend on that, I can't see any reason why it would stop working if that doesn't work

[neat sorrel]

exactly

If you need more info just ask!

[whole cloud]

2.16.151879 new PROFILING branch with PERFORMANCE binaries, v8, server and client, windows 64-bit, linux server 64-bit
- Added: DLL extension access to DirectX
- Tweaked: Reduced EXE file size
- Fixed: serverCommandAvailable was broken since v7
- Fixed: enableInfoPanelComponent could not disable CustomDisplayComponent - https://feedback.bistudio.com/T162729

If you don't want to use the Steam branch, the files are also available for alternative download here:
https://drive.google.com/drive/folders/15p9j7C2nHUt6NoVfChX4YFuqzFXzblJh
Note: There are separate Dll files that also need to be placed into Game folder.
🦺

[full nova]

i wonder if i can update the profiling without steam also downloading my new subscribed mods

the answer is nope

[whole cloud]

You can download from google drive manually and put it in

[lilac hearth]

Hi @whole cloud . With this performance build, all AA missiles stop locking air targets!

[whole cloud]

must've been in v7 already too

Which AA missiles have you tested, so I can repro it?

[whole cloud]

also still looking for that.
Some repro that isn't in CUP weapons/vehicles

[lilac hearth]

I can confirm, i used with CUP

[void badger]

Verified with Static Titan AA

[empty plover]

could it be this https://feedback.bistudio.com/T178693 that messed up the AA?

[patent dome]

Don't think that is part of Prof?

[hybrid cradle]

it is not

shouldnt be anyway

[neat sorrel]

@whole cloud We've checked and this update didn't fix the ACRE issue. We switched back to creatordlc and it works, so it must be something related to the last update. If we can help in any way just ask.

[whole cloud]

I'd like ideas on why/how ACRE could be broken. I don't know it
I already asked in ACRE's discord channel

[neat sorrel]

looking at the changelog, I can't see anything that explains it, but I'm no expert. Maybe the fix "Performance degradation when many scheduled scripts are active" interferes with heavily modded servers?

I don't really know

[plain trout]

Have you tried using an older version of TeamSpeak?

[whole cloud]

Its got nothing to do with teamspeak, if simply running profiling branch on server (which doesn't touch teamspeak at all) breasks it

[light cargo]

are you using the ptt radio bind in game, right? ts3 ptt is locally speaking

[neat sorrel]

Yes, thanks for trying to help. Same server, same PC, same mission, same settings on the client and server… everything works when the server uses the creatordlc branch but not when using profiling

[analog acorn]

ACRE uses an extension IIRC, so I was gonna say it could be the changes to allow DirectX extensions that did it, but then that was only in the latest update and ACRE broke before that 🤔

[heavy galleon]

but TFAR works, right?
One might see that as a sabotage. Almost like the person making perf has a favourite one and it's not ACRE

[hybrid cradle]

doesnt dedmen hate tfar

[whole cloud]

It doesn't use it on the server side, and only prof on server side is breaking it
Thats the biggest thing that doesn't make sense, why does serverside break it but clientside doesn't

[knotty wraith]

and what exactly needs to be changed in the manpads configs so that I can fix it myself? I am sure that the CUP authors will not do this until 2.18

[heavy vortex]

The only thing I remember ACRE doing on the server side is sending epic quantities of radio channel data :P

[whole cloud]

I don't even know what/if its broken.
That's why I want non-cup data to see if thats also bad

[fair ore]

Update: Reproduced with no mods loaded using the launch_B_Titan_F against a flying enemy helicopter

unable to lock with T

[full nova]

another report of that aswell with video
#warlords_discussion message

[inland dew]

official vanilla warlords
#warlords_discussion message

[nocturne obsidian]

since last update titan AA wont lock

i dont use mods

just proton linux

[lilac hearth]

any hotfix ?

[analog acorn]

It's 6am, on Sunday, and if you read up just a couple of messages you'll see that no one has any idea what the hell is broken. There wasn't even a vanilla repro of the lockon issue until 1700 last night. So no, no hotfix yet, and since it's Sunday it's unlikely there will be one today.

[heavy galleon]

Switch to stable branch can be a hotfix for you

[coral pumice]

whenever i try to join the official warlords server/im in the server and it's near full capacity the game keeps crashing

no mods loaded

as soon as i get past the slot selection lobby and i try joining it will load for a bit and then CTD

tried to verify game data, still crashing

this is the server

seems to be working after switching to stable

[ashen helm]

player so big that needs 10 empty slots

[hallow yew]

No crashes for us in 24 hours David

[whole cloud]

How do i get that report?

The AA launcher issue

M_Titan_AA >> Components >> SensorsManagerComponent >> Components >> IRSensorComponent

                    angleRangeHorizontal=7;
                    angleRangeVertical=4.5;

It has a lock-on range of just 4.5 degrees, from aiming point. So you need to be aiming pretty much right onto the target.
Previously that angle would just be ignored, but now that that's fixed it takes effect.
mod launchers should keep an eye on what they've set there in their config.

The "aiming direction" calculation is wrong for soldier held weapons (as opposed to vehicle weapons), it thinks the weapon is aiming straight forward, so combined with the angle, that means the target is outside of angle (unless its directly infront of you)

M_Titan_AT is even tighter at 2.3° vertical and 3.7° horizontal

[coral pumice]

apparently you dont 🥲 im checking the appdata folder but it doesnt look like an RPT was generated during any of those sessions, does profiling branch put RPTs in a different folder?

i dont have nologs enabled either so no idea why they're not there

[whole cloud]

no its same spot

[coral pumice]

then they dont exist 🥴 sorry, no idea why

the rating.log got updated but the rpts just arent there 🤷‍♂️

if i do "launch without mods and optional dlcs" from the launcher the rpt gets generated and then moved/deleted a few moments later

ok i see why i dont have RPTs, for whatever reason whenever i launch the game the RPT from the previous session gets deleted

this is in stable

and im guessing profiling does the same thing, as thats what i was using yesteday

[whole cloud]

There is a entry in config that says how many RPTs to keep

maybe you have some mod that sets that to 1 or zero?

getNumber (configfile >> "rptFileLimit")

[coral pumice]

its set to 10

i have 4342 files in the appdata folder, maybe i should delete some 👀

[fair ore]

Strange! I was looking basically dead on and spamming my lock button without even getting the icon that it's trying to lock. Did you manage to lock something yourself?

[whole cloud]

It was thinking you were aiming north straight ahead

not where you were actually aiming, so that then failed the angle calculations

[fair ore]

Oh, that'll do it lol

[whole cloud]

When you press the PTT button, the radio doesn't work. No sound, no text box on the right corner of the screen
Cannot reproduce.
Multiplayer, I can see the radio in my inventory, and the PTT button works.
There is no sound because I'm not running Teamspeak, but everything seems to work fine.
Both dedicated server and local hosted server, both fine.

I'm just running the current profiling build on steam.
Don't know what else I should do?

[neat sorrel]

When you pushed PTT, dif you see the yellow box with the name of the radio on the bottom right corner?

[whole cloud]

Didn't notice it pasted a screenshot instead of the video. yes

[neat sorrel]

Fuck. Then it’s working for you. Can it be something related to the ace settings embeeded on the mission? I’ll check later today with a mission without an embeeded config file but I’ll need to switch back to profiling

[whole cloud]

ace settings could've maybe been flaky with the v7 build that had the broken server admin check.
But you said it was still broken with v8

[neat sorrel]

I’ll check later and get back to you. Thanks!

[whole cloud]

                if (fileExists INIT_PLAYER_SERVER_SCRIPT) then // server execution (local or remote)
                {
                    if (!isServer) then
                    {
                        [[[player, didJIP], INIT_PLAYER_SERVER_SCRIPT], "bis_fnc_execvm", false, false] call bis_fnc_mp;
                    }
                    else
                    {
                        [player, didJIP] execVM INIT_PLAYER_SERVER_SCRIPT;
                    };
                    INIT_PLAYER_SERVER_SCRIPT call bis_fnc_logFormat; // log only if executed
                };

If the file exists, it does a remoteExec.

So simple fix. The link you sent https://community.bistudio.com/wiki/Event_Scripts#initPlayerServer.sqf already has a link to a alternative method.
Just use the alternative method, and give the file a different name.

What would we do to fix that issue? make it use remoteExec "execVM" instead of BIS_fnc_execvm?

[merry shuttle]

I have just started a dedicated server session (no mods and addons) and I don't use the initPlayerServer script or bis_fnc_execvm:

17:36:58 Scripting function 'bis_fnc_execvm' is not allowed to be remotely executed
17:36:58 User Xeno (XXXX) tried to remoteExec a disabled function: 'bis_fnc_execvm'

So it must be somewhere else.

[whole cloud]

Mh but how do we find out what it is

[heavy vortex]

special-case logging hook on remoteExec? :P

[neat sorrel]

@whole cloud tried again with a dummy mission and only CBA and ACRE. It works with creatordlc but it doesn't with profiling. I can send you the mission and give you access to the test server if you need

[whole cloud]

Send mission please yeah.
I did try in a otherwise empty VR mission with only CBA and ACRE

[neat sorrel]

👌

[spiral pond]

p:\a3\functions_f\initfunctions.sqf
602 [[[player,didJIP],"initPlayerServer.sqf"],"bis_fnc_execvm",false,false] call bis_fnc_mp;

p:\a3\modules_f\hc\data\scripts\hc.sqf
167 [["A3\modules_f\HC\data\scripts\HC_local.sqf"],"BIS_fnc_execVM",nil,true] call BIS_fnc_MP;

        //--- Wait until module inits are initialized
        [] call bis_fnc_initModules;
        0.75 call bis_fnc_progressloadingscreen;

        //--- Execute automatic scripts
        if (!is3DEN) then 
        {
            if (isserver) then {
                [] execvm "initServer.sqf";
                "initServer.sqf" call bis_fnc_logFormat;
            };

            //--- Run mission scripts
            if !(isDedicated) then {
                [player,didJIP] execvm "initPlayerLocal.sqf";
                [[[player,didJIP],"initPlayerServer.sqf"],"bis_fnc_execvm",false,false] call bis_fnc_mp;
                "initPlayerLocal.sqf" call bis_fnc_logFormat;
                "initPlayerServer.sqf" call bis_fnc_logFormat;
            };```

[nocturne obsidian]

not working https://files.catbox.moe/3feoxh.mp4

[whole cloud]

The first one has a fileExists check.
The second seems to be missing it! Great find! Thank you. Should be easy to put another fileExists check around that one

[whole cloud]

fileExists check won't fix the second one, the HC_local.sqf must be moved into a BIS_fnc function, and that function be remoteExec'ed directly. Which is a bit more work

[whole cloud]

The ACRE issue only happens on linux server. And there is no fix for it in todays build

[whole cloud]

Ah I forgot to post, twice

2.16.151893 new PROFILING branch with PERFORMANCE binaries, v9, server and client, windows 64-bit, linux server 64-bit
- Fixed: JIP clients did not always synchronize properly - https://feedback.bistudio.com/T171670
- Fixed: Soldier-held AA/AT weapons with lock-on were not able to lock onto targets

If you don't want to use the Steam branch, the files are also available for alternative download here:
https://drive.google.com/drive/folders/15p9j7C2nHUt6NoVfChX4YFuqzFXzblJh
Note: There are separate Dll files that also need to be placed into Game folder.

[whole cloud]

I figured out the ACRE issue..

initialization order from hell.

The getMissionConfigValue script command is defined in file1, and says its argument type is StringOrArray.
"StringOrArray" is defined in file2.

The compiler ordered the game start initializers, such that the file1 initializers run first, before file2.

file1 initializes, the command uses StringOrArray, which isn't yet initialized, the argument type is NULL.
file2 initializes later, and sets StringOrArray to a proper script type.

Script tries to use getMissionConfigValue and passes an array. The command should take StringOrArray, so array should be fine. But when the command was initialized, it stored the value of StringOrArray, which was null.
So now it checks if the provided argument value is of type null, which it isn't, so the typecheck fails and it throws script error.

It broke recently, because we had about a dozen copies of "StringOrArray" type scattered over the code, that was refactored into a single one that's just referenced by the other files.

[whole cloud]

This profiling branch has better logging for the world-end crash, or game-exit crash

[whole cloud]

ACRE fix for linux servers is coming tomorrow

[hybrid cradle]

get ninja'd

[midnight oyster]

Anyone else getting "Stack Buffer Overrun" error on Command Line processing step of game start-up from the A3 Launcher?

[whole cloud]

Do you have a crash report?

Please gib. If you're quick enough we could get it fixed today.

[whole cloud]

2.16.151900 new PROFILING branch with PERFORMANCE binaries, v10, server and client, windows 64-bit, linux server 64-bit
- Fixed: Some script commands (getMissionConfigValue, ...) could not be executed on Linux servers, breaking some mods

If you don't want to use the Steam branch, the files are also available for alternative download here:
https://drive.google.com/drive/folders/15p9j7C2nHUt6NoVfChX4YFuqzFXzblJh
Note: There are separate Dll files that also need to be placed into Game folder.
📻

[midnight oyster]

see your DMs for crash report link and related notes

[lilac hearth]

Hi @whole cloud , whats up? I have a dedicated server runing for small number players (~12), do you recomend use the performance branch or keep it on vanilla ? thanks

[whole cloud]

Doesn't matter.

[lilac hearth]

thanks

[random isle]

the more people use it the fast we make the game better

[lilac hearth]

nice, thanks !!

[trim ether]

We use Perf builds on our servers, that's a 24/7 public server and two private ops per week, for the past 4-5 years. I can only recall 2 incidents where we had an issue (picked up before the Op) and rolled back to the vanilla build. So if you have concerns over reliability, we have never experienced them. Also, as Omon XR points out, the more people that use Perf build the better the feedback we can provide to Dedmen.

[lilac hearth]

Hi @trim ether Thanks for share your experience.

[empty goblet]

q, anyone else observing dedicated server crashes in past 2 weeks ?

[knotty wraith]

every day, the server crashed once - yesterday it switched to the stable version, there are no problems

[inland dew]

no.
vanilla PvE with hundreds of AI, armor, helis, jets and 30 players, where each can also recruit several AI, with automatic restart every 6 hrs

used 151860 version, where the debug console was not working, without reverting to previous version.
also used 151879 and 151893

no problems

thankfully debug console access and AA/AT lock were fixed relatively quickly

[random isle]

i sent dump last night to dedmen of crash

[empty goblet]

please, any chance you collected at least one crashdump and reported it (FT?) on 151900 ?

[knotty wraith]

crash dump no, give me a day - I will transfer the server to 151900 - crash dump will be

[heavy galleon]

I've got a linux perf running 24/7 on newest and no crashes so far. It's not used much, it's mainly as a "test your mods and take a quick look at the mission" kind of server, but no crash

[knotty wraith]

I didn't want to write about it: there can be many reasons - mods, your code + rpt shows nothing

I didn't have to wait long)

where should I send the crashdump?

[hybrid cradle]

dedmens dms

[knotty wraith]

discord does not allow you to send files of this size

[hybrid cradle]

.zip

[analog acorn]

If it's too big even after being compressed into an archive, upload it to Google Drive or something and send the link

[knotty wraith]

tnx

[light cargo]

zip it up if it's over 25mb

[regal blaze]

i think you should change that now...

[heavy galleon]

Dedmen fix

[regal blaze]

[empty goblet]

please use feedback tracker ...

[molten berry]

So I talked in #arma3_troubleshooting about a replicable crash, but since I'm on the profiling branch, can someone test it on regular branch?
The crash occurs every time I call in a Laser Designated Artillery shell from an M4 scorcher trough the support menu (using a support requester and virtual support provider module). I call in the arty support, wait until 5 seconds before splash to laser the target, lase it and the game crashes once the round impacts. Happens in both SP and MP, tested in a blank editor mission as well. I just need someone to test it on main branch before I report it

[cloud nacelle]

Could you DM me a mission to test this in

[molten berry]

Hey @whole cloud , I just confirmed with @cloud nacelle that the crash is only on the profiling branch, do I dm you the crash dump?

[silk pewter]

rather make an FT ticket

[jagged hemlock]

Is man portable aa/at fix is about sensor range or about something else?

[light cargo]

it says right there, it's about the missiles failing to lock/target

[whole cloud]

And that missile targetting bug is also what causes the crashes now.
I found out as "sensor owner" we pass the vehicle that shot the missile.

So if a jet shoots a missile, the "sensorOwner" is the jet. And we use the jet's position and orientation for missile targetting.
Even though the sensor is in the front of the missile itself. So if you shoot a missile with a jet, and instantly turn away before the missile enters seeking phase, it will fail to target because it uses the jets orientation instead of its own.

Previously that variable was unused, so the mistake wasn't noticed.
And in some cases, on a server, the shooter of the missile is null. Of course the sensorOwner would never be null, as the sensor is on the flying missile itself. But if you pass the wrong thing, and that thing is also null it goes boom

Broken since missile targetting was added in 2018 😄

[silk pewter]

Holy…!

[light cargo]

damn, that's a tough one

[indigo anvil]

so crashes are caused by a fix that revealed an ancient dormant bug? 😄 Hey code aint broken if never called *that way * >)

[whole cloud]

That about sums up development on Arma 3 🤣

[indigo anvil]

in my experience every large older codebase. 🙂 Some serves millions of customers 😄

[whole cloud]

And trickle down issues due to the bug.
PassiveRadar sensor on missile, detects activeRadar's. But it only detects ActiveRadar's, that can target the vehicle containing the passive radar sensor.

So like if the active radar is only a forward facing cone, and the passive sensor is not in the cone, then the passive sensor also cannot receive the active radar's waves.

So previously, the code checked if the active radar, could target the jet that shot the missile (it should be checking the missile itself instead).
But the code for checking that only supports checking if vehicles can be targeted, a Missile is not a vehicle

And even if it did, active radar cannot target missiles 😄

[woven loom]

Dear god

Well if you are able to enable targeting the missile, then that might pave the way for realistic CIWS and C-RAM...

And I guess APS

[whole cloud]

I tried but its too much. All the targetting checks fail because they rely on config entries that missiles don't have.
I'll just make a workaround for passive radars on missiles and let it be at that.

[tribal pasture]

Shooting down MLRS missiles could've been a great use case too

10 years too late I guess

[plain trout]

We had massive lag issues/freezes with I&A when the AMRAAMs were first introduced.

Any chance to activate that missile targeting feature so mods could implement that into their configs?

[whole cloud]

As I said I won't implement the ability to target missiles. That would be a whole new feature that's not planned

[silk pewter]

don't missiles need to lock (using the plane) then once fired, use their own sensors? it was using the vehicle on both?
(also, some missiles are "dumb" and only use the original target/marking, is that covered too?)

[whole cloud]

I don't understand the question

I'm talking about passive-radar missile that self-seeks its target after it was fired, with its own sensor

[silk pewter]

but for these, should the (pre-fire) lock happen with the plane's sensors?

anyway, if it's fixeded, it's gr8, gr16even

[whole cloud]

There is no pre-fire lock

It is fixeded now..
But same as in my last tests, the HARM missile from the black wasp, never hits its target.
It always goes over. But I didn't touch that and could reproduce that being that way in the old Arma versions too

Seems broken to me, but don't want to touch that also without a proper ticket about it..

There is more questionable code.

There is "are our active radar emissions revealing our radar to the enemy"
aka do we notify the enemy that we are here. And, it only notifies enemies that are target-able. But that doesn't make sense to me.
A radar sensor will not sense any radar, if it cannot be targeted.

A radar sensor in a AA vehicle senses radar.
A radar sensor in a parachute, doesn't sense radar. (Because parachute has config radarTarget=0;)

[tribal pasture]

Thick 3d lines

[whole cloud]

huh

the thing drawLine3D uses, has a width parameter. But its just left at default.
So we could add width parameter to the command

interesting. Theoretically one would only need to make a FT ticket to request addition of that parameter

Default width is 3, these lines for shots diag are 10

[worldly badge]

I do believe we already have more than one tickets

https://feedback.bistudio.com/T122512
...One, actually

[whole cloud]

v10 performance + crash fix for the missile targetting.
Don't have time to do a proper release this week.

Otherwise you'd have to go back to v6 to get rid of the crash

The proper release probably wednesday

[light cargo]

can't afford for data changes?

[whole cloud]

just too much work in general

[empty goblet]

can those config entries be added via patch (server side addon) ?

[whole cloud]

There are no config entries to be added

[empty goblet]

All the targetting checks fail because they rely on config entries that missiles don't have.
guess i misunderstood it

[restive pilot]

Harem missile?

[light cargo]

yeah basically, the config data could be added but it's going to need a code change along with it

and perf/prof doesn't make breaking code/data changes to stay compatible with stable

[heavy galleon]

ACE will have it one day it's looking really good

[knotty wraith]

new crash 🤷‍♂️

[empty goblet]

gotta dig that dump 😁 it's binary

[woven loom]

hmm we were playing Force Recon on a linux DS, everyone including the (stable) server crashed at the end of the mission 🤔

i crashed on profiling (steam)

Access violation

[light cargo]

might be worthy to send the mission and the crash dumps?

[woven loom]

It was GM force recon but on cham

i crashed again, just profiling client, on successful mission completion when the replay plays

same access violation

have crash dumps for both, and yeah can send the mission too

[uncut linden]

Sorry to hear that. But in the future, do you think you could check the AI behaviour in tracked vehicles? I noticed that after spawning and getting their first waypoint assigned AI controls wheeled vehicles and boats just fine, but for tracked vehicles it does a lot of wiggling left-right at place. My theory is that is because the tracked vehs have "swapped" left-right controls in reverse mode. Think you could check that in future versions? Thanks.

[heavy vortex]

You mean when you give them a waypoint behind them?

[plain trout]

You can try to disable aisteeringcomponent. It was discussed here some time ago.

https://community.bistudio.com/wiki/useAISteeringComponent

[fallow mason]

This problem has been solved? i hosted a game lastnight and the server crashed and it never reached more than 4gb

https://cdn.discordapp.com/attachments/893628928160329788/1254478269647159450/image.png?ex=6679a36e&is=667851ee&hm=5794a78e92e534889d36fb9fd86c4c51d1794eabe2df8572648917060b6f7921&

Im using linux live server( not perf) and with -hugepages -bandwidthAlg=2 -maxMem=10239 params

[knotty wraith]

there was no more crash, maybe it was an accident

[whole cloud]

Its time for me to look at crashes. And I have no crashdumps in my inbox 😢

[whole cloud]

No changes to AI behaviour.

[whole cloud]

yes. Linux still has a memory limit, I forgot how much it was. But it was extended

[whole cloud]

yeah I bunged that up on friday evening 😄
That only included half of the fix
Thank you

[knotty wraith]

I had no doubt - you are the only one who still saves this game!

[heavy galleon]

And then breaks it*

insert dedmen fix gif

[indigo anvil]

any chance for a full fix rls? We could boot that up on a test server

[whole cloud]

tmrrw

[woven loom]

I'll send today! ❤️

[woven loom]

sent!

[whole cloud]

2.16.151911 new PROFILING branch with PERFORMANCE binaries, v11, server and client, windows 64-bit, linux server 64-bit
- Fixed: Server crash during missile targeting
- Fixed: Possible cause for BattlEye query timeouts when having high packet loss

If you don't want to use the Steam branch, the files are also available for alternative download here:
https://drive.google.com/drive/folders/15p9j7C2nHUt6NoVfChX4YFuqzFXzblJh
Note: There are separate Dll files that also need to be placed into Game folder.
🫠

[knotty wraith]

HC how does exe use server or client? I understand that it runs on the server, but you never know...

[light cargo]

it can use either, same with ds

iirc

[whole cloud]

HC how does exe use server or client?
You mean whether HC is server exe? yes HC is server exe

[light cargo]

it runs separately but via server and client exes

[whole cloud]

client exe afaik is not intended to be used that way

[light cargo]

but it can, why is it present if is it not intended/supported?

[whole cloud]

Arma clients kinda need to be able to client. You can't remove that, that's how playing Arma works.

[light cargo]

i'm talking about why the ds and hc are in the regular exe

[whole cloud]

Because client has local-hosted server ability. So it needs the hosting ability

[empty goblet]

because in A2 hc was special client with server as addition ... in A3 it was decided to move HC to server binary for security reasons and to make it streamlined .... the hc alike client is also useful for testing

[light cargo]

interesting

[light cargo]

separate exe for hc or bundled in dedicated? i forget, a2 didn't have ds on client, right?

so hc depends on the dedicated server infrastructure?

[empty goblet]

always had ds on client (literally evolved from hosted coop w/o ds) ... but the way how HC was done in A2 was insecure and inefficient

[light cargo]

i see, is a3 hc much better? thanks!

[indigo anvil]

I've just got a client crash during EU11 gameplay. Where should I send it?

[whole cloud]

👋

[light cargo]

is it on perf/prof? send to dedmen dm

[indigo anvil]

DM sent with a link

[upbeat thistle]

Titan Lock (AA and MPRL) and also Tigris is broken on profiling version

you cant lock vehicles

[rancid coyote]

Same, I can't actually find any AA missles that work in profiling branch. I even disabled all mods as thought it might be those, but definitely a profiling issue.

[upbeat thistle]

Yeah we can’t kill helicopters camping the skies like that

[orchid mesa]

Same with the RHS mods. The M6A2 can't lock 99% of the time, and when it does, it's only for a second before losing it again. The handheld Igla isn't quite as broken, but still barely works - if you lock a vehicle with it then fire, the missile goes off into the sky and never tracks its lock

[knotty wraith]

The handheld Igla isn't quite as broken, but still barely works - if you lock a vehicle with it then fire, the missile goes off into the sky and never tracks its lock

I confirm

[weak panther]

+, same on dev build

[upbeat thistle]

Needs to be fixed

[silk pewter]

no no, let's keep things broken of course

[fair ore]

obviously telling people the obvious (that bugs should be fixed) is productive

it never comes across as entitled and annoying

[heavy galleon]

Hey, they are a general!

[light cargo]

jeez

[empty goblet]

85.190.148.52:2102
KOTH stress-test at server-side

100 players, 250 fps
120 players, 200 fps
130 players, 150 fps
140 players, 130 fps 
150 players, 100 fps (previous run 120)
160 players,  90 fps (previous run 110)
170 players,  80 fps (previous run 100)
175 players,  60 fps 

[tribal pasture]

15 FPS on clients

[empty goblet]

ye now that NEEDs some optimization 😁

[knotty wraith]

there has always been a Tushino project since Arma 2 where 200+ players played... It is very sad that BI has not found the time to adopt this unique experience.

[tribal pasture]

Works good in SP

[empty goblet]

well, now those can be recreated in A3 (mostly, the client performance is still issue)

[knotty wraith]

have you ever had any options to make an HC on the client side? thus, to swing the processor to at least two parallel running cores to the maximum

HC on the server side has completed its task in filling the cores to the maximum - this is a real result!

[empty goblet]

that's not where the bottleneck is ... wouldn't help you on client much or at all

[knotty wraith]

by the way, is the alternate upload to the server as in Dayz essential?

[empty goblet]

what you mean by 'alternate upload' ?

[knotty wraith]

the client is loaded in the queue - one by one

[empty goblet]

that would be good question to ask dedmen , if there can be something done about

[heavy galleon]

Not sure what he's on about, but Tushino is on A3 for a long time. So are other (large scale Russian) PvP events.
You could have asked me, I run PvP events myself and last time we had 180+ players anf FPS were in the 300reds.
When Russians play event vs event, it is often at 220 players and the FPS are fine.
For instance WOG vs Tushino on WOG server (not the best HW tbh), FPS part of the stats - https://en.stats.wogames.info/games/3275/
(I can't send pics here, but the FPS graph from the stats page is here too if you don't want to open that - https://ctrlv.cz/R1Oy )

[knotty wraith]

I said that Tushino has a unique 200+ game experience since arma2!

and so it is in arma 3

[empty goblet]

what i meant was the scale of the 200-250 players with AI ... for long time that was really hard to recreate reliably compared to the custom overoptimized A2 servers

[empty goblet]

but if you doing it for 'some time' i'm just happier ... tho nobody told me that 'good news' 😁

{want to squeeze more client-side FPS anyway , just as server-side}

[heavy galleon]

Well, no AI. It's PvP only. But quite some time now yes 😄

[knotty wraith]

it feels like they are hearing about Tushino BI for the first time

[empty goblet]

ye i was more about the combined op ones ... i played several with around 200 players and like 200 AI in A2CO times and it was enjoyable but they spent like 3 years perfecting the performance

[heavy galleon]

Russians spent 10 on A3 by now 😄

There's number of PvP projects that get large numbers. Pretty enjoyable too.
Too bad it's not as well developed in English speaking countries.
(I am trying though)

[empty goblet]

well A3 got massive amount of optimizing since release so now it's very close to performance similar to A2 (take in mind in A2, the whole group of AI just used the leader's (both squad and group) brain, while in A3, each AI under leader has full brain not just leader)

that's why early on people were like uh oh i can't run my 1500 AI scenario which worked in A2CO and now runs at 5fps in A3

[heavy galleon]

Yeah. I am happy for any improvement

Especially stability. When the server crashed on my last event with 190 people, it was... Bad press let's say

[empty goblet]

ye w/o doubt freeze or crash or stall (mass connection drop) are bad on large servers ...

but that's unfixable (except perfect stability/performance) until fully server-centric and add some multi-server meshing

[heavy galleon]

So, 2026 you say?

[empty goblet]

i don't want to give promises, tho i would like all the server-goodies StarCitizen has for A4 and beyond 😁 {one Arma veteran can dream right?}

[heavy galleon]

Go hire some talent
I heard uni just ended and there's a lot of skilled people out there

[empty goblet]

if it was that simple, we would be already rolling out ArmaVerse 😁

[woven loom]

Hopefully Arma 4 will release before Star Citizen 😛

[spiral pond]

better means for performance profiling and especially logging+analysis available to the community would go a long way to identify perf issues and bottlenecks

[median belfry]

What I did for a3s was hiding and turning off simulation of each player not in their view distance client sided. Same for ai and zombies. And holy the fps were going up when there were a lot of players on the server.

But if players are not really spread out on the map and all fighting on one point it will not help

[heavy vortex]

yeah I don't understand why non-local simulation runs for non-visible units.

[median belfry]

// Parameters
_p = player;
_all = (AllUnits) - [_p];   
if (_all isEqualTo [] OR !(isNil "A3S_var_SpectateEvent")) exitWith {};

//Unhide to get the position synced 
{_x enableSimulation true;_p reveal _x;uisleep 0.02;} count _all;  
uiSleep 0.5;   
_ViewD = (viewdistance max 500) + 100;         
_near = _all select {(_x distance2D _p) < _ViewD};  
         
// Hide the units which are not around
{
    _U = _x;
    _V = vehicle _U isEqualTo _U;
    if ((!(_x in _near) OR ((getPosATL _U)#2) > 80) && _V) then {   
        _U enableSimulation false;
        _U hideObject true;   
    } else { 
        _U hideObject false;
    };   
    uisleep 0.02;  
} count _all;   

people can try it out , but make sure to run it in a loop and check it every minute

[heavy galleon]

I see sleep and I cry

[median belfry]

why

[heavy galleon]

Scheduler

[median belfry]

yea I made it for scheduled enviroment

otherwise the position will not update

since you disable the simulation of the player will freeze him on the position and if you dont give time to update the position after enabling the simulation for a short time it will just thinks he is still there

[heavy galleon]

cba_fnc_executenextframe or something like that

[silk pewter]

not everyone has CBA :p

[median belfry]

also some calculation cycles you dont wanna run in unschedulded enviroment since it can causes lag spikes if you dont have breaks in between

[heavy galleon]

That`s just skill issue then

[heavy galleon]

Sounds like great usecase for an extension

[median belfry]

if it just wouldnt take weeks to get approved by battleye yea thats the best solution

[tribal pasture]

I used to do this simulation disable a lot in Wasteland. Mostly worked fine except some issues with vehicles like you could have engine on sound globally

[whole cloud]

AA missile lock not working.
When a missile is spawned, it does an initial check whether it can see the target (that was given to it at spawn, by the shooting device that selected the target)
During that initial check, the missile isn't fully initialized yet and its position is 0,0,0. So the target is out of range. So it cannot lock, so it discards the target.

[empty goblet]

one must wonder what else gets init on 0,0,0 (or worse somewhere below ground) and could fail that way 🤔

[whole cloud]

I reverted most of the changes now and just made only the missile Seeking phase check whether target is in range/angles.
Hopefully that's better

[woven loom]

Why is it initialised at 0,0,0 when it's actually on the pylon of the aircraft?

[light cargo]

the plot thickens

[woven loom]

Just wondering if it can't inherit the position of the vehicle firing it?

[whole cloud]

The position is set after the entity is spawned

Immediately after, it gets the position

2.16.151931 new PROFILING branch with PERFORMANCE binaries, v12, server and client, windows 64-bit, linux server 64-bit
- Tweaked: Handling of Eden Editor mouse placement (can now place on top of all objects and upside down and on the sides of objects; please report issues that you find with this)
- Tweaked: Face command now properly returns the current face as class name - https://feedback.bistudio.com/T85502
- Fixed: setFace was not setting a custom face properly - https://feedback.bistudio.com/T85502
- Fixed: Blank face after a player with a custom face leaves the server - https://feedback.bistudio.com/T85502
- Fixed: Face command not always returning the set face - https://feedback.bistudio.com/T85502
- Fixed: Crash when cutText replaces itself during "Unload" Event Handler - https://feedback.bistudio.com/T182375
- Fixed: AA/AT missiles would loose lock after launch

If you don't want to use the Steam branch, the files are also available for alternative download here:
https://drive.google.com/drive/folders/15p9j7C2nHUt6NoVfChX4YFuqzFXzblJh
Note: There are separate Dll files that also need to be placed into Game folder.
🚀

[upbeat thistle]

Infistar menu is not working on normal version

i was using normal version since yesterday night. I opened the game again today without profiling version and now infistar menu is not working

[whole cloud]

¯_(ツ)_/¯

Talk to infistar people and find out why

[upbeat thistle]

ok

[whole cloud]

I have no way to test that or guess what might be wrong with it

[upbeat thistle]

ok mate

its ok

it works i opened the game twice and i works now

so all good

🙂

[knotty wraith]

is it necessary to update it every time RTDynamics_64.dll ?

[hybrid cradle]

yes

its not in stable yet

[knotty wraith]

I'll ask a different question: 11 RTDynamics_64.dll is it different from 12? ?

[whole cloud]

no

[restive pilot]

Every time? I think it only needs to be updated once (due to C++ RT change after compiler update)

[shell lion]

New perf branch has issues with GRAD Trenches.
If you want to build a trench, the placing doesn't match anymore.
It's only on PERF branch and only on this version.
So i don't know what kind of changing it effects (mayby the Handling of Eden Editor mouse placement).

Is the change correct, or is it a mistake ?

If it's now correct then I inform GRAD Trenches Mod Team.

[light cargo]

maybe check what they think about it

[lilac hearth]

hey hey, I don't know if there is any awareness or incompatibility with a mod that my server is using, but it happened a few times in the PERFORMANCE version that some players became invisible, has anyone had any experience like this?

[light cargo]

maybe check the rpt and do a binary search and try to reproduce without mods. this is a known issue that happens in many scenarios and it really varies as of when and why

[lilac hearth]

thx

[light cargo]

fortunately its often a rare issue

[lilac hearth]

good to know. Do you know if theres any relation with performance build or not ?

[hybrid cradle]

no

[lilac hearth]

thx

[whole cloud]

I can't think of any change.
It was not broken in previous version?

The eden change really is literally only when you move your mouse around in eden editor, nothing else.

[whole cloud]

I can reproduce the issue

The mouse controls also don't work, cannot cancel or confirm digging

[whole cloud]

I found it.
Its a bug in a SQFC (Script bytecode) compiler (Not ArmaScriptCompiler, that one is fine)
It was passing the wrong type of parameters. In latest profiling a error message for it was added, and it was hotfixed to handle the wrong parameters.

Turns out, besides it passing the wrong type of parameter. The value inside that parameter is also wrong. Now breaking all mods that were built using that faulty compiler.
They were already passing the wrong stuff previously, so they were always broken in some way. But apparently it was lucky that it worked anyway.

[quaint flame]

So Hemtt should be fine now or are there still things to be done on their end?

[whole cloud]

I don't know if HEMTT released the latest version where it uses offsets instead of strings. Its fixed internally but dunno if released.
Mods would need to be repacked with it.
And if HEMTT previously passed just empty strings causing it to be broken, then I'm not sure if new HEMTT with offsets wouldn't make the same mistake.
Seems latest HEMTT breaks ACE interactions, so probably still not fine.

[whole cloud]

For the technical.

HEMTT (It seems a quite old version) wrongly encoded the content of code blocks, as empty string.
Every code block stores the instructions to be executed, and the string so you can str or toString it back to read what's inside it.
I thought it wouldn't matter when its broken and you do not str/toString the code. But it does.
isEqualTo on Code, compares the string, not the instructions.

So. HEMTT creates code blocks with empty string, even though they aren't empty.
isEqualTo only compares the strings.
And CBA...

https://github.com/CBATeam/CBA_A3/blob/master/addons/common/fnc_addPerFrameHandler.sqf#L32C25-L32C27
Skips adding your PFH, if your code block isEqualTo {}. Which means if its toString _code == ""
And because the code block was wrongly set to be empty string.. CBA skips all the PFH's you try to add.
Including the PFH that GRAD Trenches uses for trench placement.

Encoding content of code blocks as strings wasn't even supposed to be done and was broken previously.
That bug was fixed after troubleshooting issues with HEMTT's SQFC compiler and finding this.
But I didn't expect that besides it encoding strings (where its supposed to be offsets), it was also encoding wrong strings.

GRAD Trenches needs to be repacked either without SQFC, or with ArmaScriptCompiler SQFC, or maybe with latest HEMTT version (if that was even released yet, don't know).
They already know about it.

[spiral pond]

is this about the above prob?

SQFC Code block string uses constant index instead of offset encoding. This is only supported with the main code block. Tried to use constant index: 0

or this

Error in expression <;

    cba_help_DiaryRecordKeys = _unit createDiaryRecord ["cba_help_docs", [loc>
  Error position: <createDiaryRecord ["cba_help_docs", [loc>
  Error Foreign error: Unknown enum value: ""
File /x/cba/addons/help/XEH_postInit.sqf..., line 298
 ➥ Context:     [] L245 (/x/cba/addons/common/XEH_postInit.sqf)
    [] L66 (x\cba\addons\common\init_perFrameHandler.sqf)
    [] L65 (x\cba\addons\common\init_perFrameHandler.sqf)
    [] L300 (/x/cba/addons/help/XEH_postInit.sqf)```

[whole cloud]

yes

[whole cloud]

No. That's different. We never released that problem. Assume you're on internal build. That's a WIP thing

[empty goblet]

hello, anyone has new crashes from weekend sessions on performance or RC of hotfix #2 binaries ?
please PM me the downloads if you do, thanks

[charred lava]

Thermal vison (white and black) is oscillating between getting completely washed out vs seeing fine details on surfaces, takes about 20 seconds for a full cycle - is that normal?

[light cargo]

i think i've seen that in stable too, it really depends on your environment and how it affects the exposure or something

[azure jewel]

Bug found: Can not control UAV on its passenger seat

Reproduce:

1. Place UAV operator
2. Place UGV Stomper
3. Play as the unit in step 1
4. Get in the vehicle in step 2
5. Connect terminal to UAV
6. Take UAV controls
7. Nothing happens

stable branch is fine

[feral harness]

will try to push fix for the next profiling

[fast hornet]

v10 and v11 both crash about once per week for us btw.. v6/151824 runs stable.

The last two times it happened we had nothing useful in the logs, but this time on the latest crash, about 30mins ago, v11 managed to write bidmp and mdmp files.
@whole cloud you want them? 🙂

[indigo anvil]

he does a 100%. send dm in link

[fast hornet]

DM sent.

[whole cloud]

2.16.151989 new PROFILING branch with PERFORMANCE binaries, v13, server and client, windows 64-bit, linux server 64-bit
- Tweaked: Zeus height-only movement now blocks horizontal movement
- Tweaked: Improved format script command performance - https://feedback.bistudio.com/T81725
- Changed: -debug parameter on Dedicated Server while on Profiling branch now forces full-dumps on crashing
- Fixed: Localized title and descriptions for exported missions were not showing in the scenarios menu - https://feedback.bistudio.com/T179398
- Fixed: "Killed" Event Handler was not triggering for upside down helicopter crashes - https://feedback.bistudio.com/T181306
- Fixed: Missing falling sound for mortar shells in some cases - https://feedback.bistudio.com/T159917
- Fixed: UGV remote control from its passenger seat stopped working
- Fixed: Delay in transferring UAV remote control to a player - https://feedback.bistudio.com/T182587
- Fixed: Entity leak after remote removal of a vest or uniform - https://feedback.bistudio.com/T182649

Note: Please report if you encounter new/unexpected "Enum" script errors.

If you don't want to use the Steam branch, the files are also available for alternative download here:
https://drive.google.com/drive/folders/15p9j7C2nHUt6NoVfChX4YFuqzFXzblJh
Note: There are separate Dll files that also need to be placed into Game folder.
🧹

[fast hornet]

Would -debug have any performance implications (while not crashing :D)?

[whole cloud]

Yes. debug has extra debug logging

[knotty wraith]

this appeared after the update

15:31:20 Error in expression <tRain 0; 
1 setfog 0; 
1 setOvercast 0;
simulWeatherSync;
{
private _pos = getPo>
15:31:20   Error position: <simulWeatherSync;
{
private _pos = getPo>
15:31:20   Error Undefined variable in expression: simulweathersync

code

        1 setRain 0; 
    1 setfog 0; 
    1 setOvercast 0;
    simulWeatherSync;
```before the update, it always worked without errors

[whole cloud]

Its fine in dev branch. Might be a failed merge. Checking..

[whole cloud]

In your server logs right?

[knotty wraith]

yes

[whole cloud]

Ok yeah I got it. It might stay till friday. I think it shouldn't break much, the command does nothing on server anyways

[empty goblet]

q, is the fulldump change for -debug w/o the confirm dialog ?

[whole cloud]

yes

I need a fulldump from a server owner who doesn't have access to UI. So on server the confirmation dialog is skipped

[empty goblet]

ye, breaks any mission where weather sync is forced on clients which update to this version

[whole cloud]

The script command works fine on clients. It only doesn't work on servers.
And on servers, the script command does literally nothing

[empty goblet]

i'm not talking about server ... the properly made mission / server shall RE it to clients

[whole cloud]

I would expect RE to still work

[empty goblet]

the problem is lot of the code executes on everything including server so likely will error out on server too

[whole cloud]

Yeah it'll spam errors for a bit

Not great not terrible

[empty goblet]

ok, when you will be fixing it ...
maybe allow option to be able define keyframes period for simulWeather
numKeyframesPerDay = 48; from class SimulWeather
enforced from server-side cfg (as another entry for class AdvancedOptions ) and/or description.ext
instead of it being just hardcoded value (of 24 hours divided by 48 thus each 30 gametime minutes is one key-frame)

[whole cloud]

It might stay till friday.
I have already fixed it. But a build today probably won't happen

[spiral pond]

question about this:

Revision: 151932

You also need to prefix description.ext briefingName and overviewText with @ not $ if you use description.ext over 3DEN, see
https://community.bistudio.com/wiki/Description.ext#briefingName
https://community.bistudio.com/wiki/Description.ext#overviewText

so is both @ and $ supported in desc.ext going forward? or what happens with missions using $ so far?

my understanding was that @ was sqm exclusive

in addition you need to define both in sqm to appear in the MP lobby (dunno if Eden does the same/writes to sqm)

another one on:

• Changed: -debug parameter on Dedicated Server while on Profiling branch now forces full-dumps on crashing
  is this using the same limit set in main config as for number of rpt files?
  plus full dumps are easily couple GB large, right? so 10 * x GB might turn out problematic

[whole cloud]

Looking at code, displayName was always checking for @
and overviewText was never localized before, so there shouldn't be old missions with $

[whole cloud]

yes. mdmps and rpt files use that limit.
Yes they are very large. -debug parameter is intended to be used for development.
In a development environment, big crashdumps are fine

[spiral pond]

alright. i tend to agree - just a warning/info message should be put out for server admins and hosts to be/become aware with 2.18 i'd recommend

p:\a3\missions_f_gamma\mpscenarios\mp_coop_m03.altis\description.ext
15 briefingName = $STR_A3_MP_COOP_m03_briefingName;
16 author = $STR_A3_Bohemia_Interactive;
18 overviewText = $STR_A3_MP_COOP_m03_overview;
19 onLoadName = $STR_A3_MP_COOP_m03_briefingName_short;
20 onLoadMission = $STR_A3_MP_COOP_m03_overview;

maybe i am misunderstanding the change or the context

or did none of that ever work and it always fell back to English?

or does it relate to this actually? (defined in sqm)

WW2\SPE_Missions_p\GameModesBI_p\WL_16_SPE_Warlords.SPE_Normandy: string @STR_SPE_Missions_GameModesBI_WL_16_SPE_Warlords_briefingName_Normandy cannot be localized client-side - move to global stringtable
and
[03] Der Marsch von Quibou nach Dangy: string @STR_SPE_Missions_Campaigns_SPE_Operation_Cobra_SPE_M03_Quibou_Dangy_March_overviewText cannot be localized client-side - move to global stringtable

[whole cloud]

"with 2.18"
This won't be in stable builds

[whole cloud]

It is different for addons

Mission from campaign addon mod. Is different from mission from mpmission pbo, is different from mission from mpmissions folder

the mod, loads its stringtable at game start, and the strings are present when the config parser reads the description.ext

in mpmissions pbo, the stringtable is not loaded when description.ext is read, it would fail to localize the strings.
The mission pbo, reads description.ext and mission.sqm first, THEN after that, it temporarily loads the stringtable from mission pbo, and manually tries to localize the @ strings

The simulweather commands issue was even super funky before this change.

The nular command simulWeatherSync is available on client and server. On server it just does nothing.
But the functions https://community.bistudio.com/wiki/setSimulWeatherLayers
Were already missing on dedicated servers, and it wasn't documented. And functions are worse because they would actually make whole script functions fail to compile if you use them.. What a mess.
And now that I fixed it, the profiling build fails, because one of the functions that were always missing, doesn't even have a dummy implementation

[spiral pond]

yeah i am aware transfer missions are different.

however at least the warning is not always right - above sample is from a mod/addon mission. aka at game start the engine parses the xml, and thus can be localized
i guess the parser goes by foldername with NAME.TERAIN\mission.sqm - regardless what mission setup it is

anyhow still unsure what to make of the change - steam downloaded and server download missions may need to be treated differently, but beyond that the existing $ setup should remain supported, no?

[whole cloud]

$ will work. Because $ is handled and resolved by the config itself.

Its like old code was
overviewText = getText (missionConfig >> "overviewText") // getText does localization automatically
New code is

overviewText = getText (missionConfig >> "overviewText"); // getText does localization automatically, if it starts with $STR. This would fail if its a mission.pbo with included stringtable. On mod missions, it still succeeds and returns localized text.
if (overviewText startswith "@") then {overviewText = localize (overviewText select [1])}; // If it starts with @, cut off the @ and localize it.

[spiral pond]

@silk pewter we will need your docu magic skills please

[heavy galleon]

What about the Addon/addon linux thingy, given the pbo does not matter anymore?

Do you need a FT for that?

[gentle canopy]

yo is this like the tech support channel?

[heavy galleon]

nope, try #arma3_troubleshooting or #arma3_questions depending on what you need help with

[gentle canopy]

thanks

[rain moth]

well, maybe it could support those with an s at the end 😉

[heavy galleon]

oh you… I’m in vacation mode ok?

[whole cloud]

I don't want one

[heavy galleon]

But would that help you in remembering?

[whole cloud]

I don't want to remember

[heavy galleon]

The pbo names are supported. The Addons some mod Devs put in just fucks it all up again

[whole cloud]

Jup

[heavy galleon]

Is it that hard?

[whole cloud]

No

Probably not

Maybe

[heavy galleon]

Can we expect it on Friday?

[whole cloud]

[feral harness]

@spiral pond specific to exported missions and mission stringtable. if you have mod stringtable it would probably work fine but mission stringtable is temp loaded in scenarios menu and if you have $ it tries to resolve it during parsing but since mission stringtable is not loaded it cant find localisation

[spiral pond]

copy. will test with new version just to make sure our setup still works as before 🙏

[real creek]

It seems a mod stopped working with perf/profiling. Is it better to report to the mod dev in that case?

[heavy vortex]

Report to both IMO.

If perf branch breaks a mod then that's usually unintentional.

[real creek]

Oh I just realized the "play music" zeus module is not related to any mods. It's breaking like this right now, as well as "play sound" https://i.imgur.com/RddpluR.png
https://feedback.bistudio.com/T182789

[weak panther]

seems broken Zeus list box(tree box), can be reproed when use any module with

[kindred radish]

My simple statusbar also stopped working without changing anything
https://pastebin.com/PjR6d9yL

Edit: Something broken with parseText format

_test = parseText format[
"
<t shadow='1' shadowColor='#000000' size='1.1' color='%7'><img size='1.1'  shadowColor='#000000' image='a3\ui_f\data\gui\rsc\rscdisplayegspectator\fps.paa' color='%5'/> %1  </t>
<t shadow='1' shadowColor='#000000' size='1.1' color='%7'><img size='1.1'  shadowColor='#000000' image='a3\ui_f\data\gui\rsc\rscdisplayegspectator\fps.paa' color='%5'/> %2:%3  </t>
<t shadow='1' shadowColor='#000000' size='1.1' color='%7'><img size='1.1'  shadowColor='#000000' image='a3\ui_f\data\gui\rsc\rscdisplayegspectator\fps.paa' color='%5'/> %4</t>",   
count allPlayers,   
5,
10,
60,     
parseText "#FFFFFF"
];
_test

Returns only 1 as output .

[spring harbor]

and here i was thinking it's just me, especially when it randomly broke on the same day

[whole cloud]

Ayayay.. thanks peeps, getting right on it!

[whole cloud]

I get 0

Format returns

" 
<t shadow='1' shadowColor='#000000' size='1.1' color='><img size='1.1'  shadowColor='#000000' image='a3\ui_f\data\gui\rsc\rscdisplayegspectator\fps.paa' color='#FFFFFF'/> 0  </t> 
<t shadow='1' shadowColor='#000000' size='1.1' color='><img size='1.1'  shadowColor='#000000' image='a3\ui_f\data\gui\rsc\rscdisplayegspectator\fps.paa' color='#FFFFFF'/> 5:10  </t> 
<t shadow='1' shadowColor='#000000' size='1.1' color='><img size='1.1'  shadowColor='#000000' image='a3\ui_f\data\gui\rsc\rscdisplayegspectator\fps.paa' color='#FFFFFF'/> 60</t>"

Which looks... alright, so parseText is having issues...

10:18:57 Wrong color format ><img size=
10:18:57 Unknown attribute 1.1'
It also throws errors

Bah we have 4 copies of the same code

Ah nevermind the format is wrong

color='%7'> turned into color='>

[whole cloud]

Your code doesn't make sense.
You have 5 arguments. But you use %7 in the format string? There is no 7

Clearly a game bug handling it wrong, but also your format is incorrect

format ["'%2'", 1] returns "'" should be "''"
Its cutting off one character, if the argument wasn't found

The zeus is probably same issue.
That _fnc_scriptName header is built using format. There looks to be a missing ' there too.
But I cannot find a bug in our initFunctions format

Because that is CBA's header
https://github.com/CBATeam/CBA_A3/blob/eedb28607471436b72da4e4f8cfe31cf2800eca1/include/a3/functions_f/gui/fn_initDisplay.sqf#L141

But their format string should be right, somehow its also missing the quote there

There you go. You'll have to deal with the errors then @kindred radish

[whole cloud]

Actually it doesn't seem to be CBA's header.

According to the script error, the header is built using
format ["scriptName '%1_%2'; _fnc_scriptName = '%1'; ", _scriptName]
with the %2 argument missing.
I couldn't find that in CBA's code nor in ACE.

Found it.

functions_f_curator/Curator/fn_initCuratorAttribute.sqf
_fncFile = format ["scriptname '%1_%2'; _fnc_scriptName = '%1';",_class] + _fncFile;

That is clearly broken, and now we throw script errors everytime we compile zeus modules scripts. crap

[whole cloud]

nice

[kindred radish]

I changed this script years ago and never saw the error nor did it error in the rpt etc
Also worked like intended since i wanted the text to be white.
#Betriebsblind as we say in germany.

[quaint flame]

Good that it throws an error now.

[whole cloud]

2.16.152001 new PROFILING branch with PERFORMANCE binaries, v14, server and client, windows 64-bit, linux server 64-bit
- Fixed: format script command incorrectly handling missing arguments, now it throws an error (but also handles it correctly)
- Fixed: Missing simulWeather script commands on Dedicated Servers

Note: It's known that opening Zeus modules throws a script error; this will have to stay until the 2.18 release. The modules still work and the specific error can be ignored.
Also please still stay on the lookout for Enum errors.

If you don't want to use the Steam branch, the files are also available for alternative download here:
https://drive.google.com/drive/folders/15p9j7C2nHUt6NoVfChX4YFuqzFXzblJh
Note: There are separate Dll files that also need to be placed into Game folder.
🍠

[quaint flame]

Seems I fell victim to the wrong argument count for format a few times as well 😄

@spiral pond For awareness, SPE has a few as well

[quaint flame]

@whole cloud a3\ui_f\scripts\gui\rscdisplaydebriefing.sqf line 45 throws format error now. Should we collect these in a ticket?

Oh shoot. This one might actually be caused by modded content

18:03:22 "BIS_fnc_log: [BIS_fnc_codePerformance] ----------------------------------"
18:03:22 "BIS_fnc_log: [BIS_fnc_codePerformance] Test Start. Code: systemChat ""test"""
18:03:22 "BIS_fnc_log: [BIS_fnc_codePerformance] Test Cycles: 10000 / 10000"
18:03:22 "BIS_fnc_log: [BIS_fnc_codePerformance] Test End. Result: 0.00025296 ms"
18:03:22 "BIS_fnc_log: [BIS_fnc_codePerformance] ----------------------------------"
18:03:22 Error in expression <        format ["<t color='#99ffffff'>%1</t>",format [localize "str_a3_cfgnotification>
18:03:22   Error position: <format [localize "str_a3_cfgnotification>
18:03:22   Error 0 elements provided, 1 expected
18:03:22 File /temp/bin/A3/Functions_F/Debug/fn_codePerformance.sqf..., line 56

[whole cloud]

Ah fun.
I already fixed the zeus one today.
Yes please collect in a ticket

[cold vale]

The fix for soundFakeFall is great, would it be possible to expand more config control over the behavior if there is anything else in-engine that could be opened up to be tweaked (for ex. adjusting when the check starts playing the sound)

[whole cloud]

How do we know what people would like to be opened up?
Anything else. Well everything, we could move all of it to script but that'd be bad.
Any examples? How would you adjust the check?

[quaint flame]

https://feedback.bistudio.com/T182822

[feral harness]

it plays approx 5 seconds before the surface, but it is better to have longer sound. this is fixed, unfixing it could break things and a bit of work as well

[tribal pasture]

Will it error for more arguments than % numbers in the string too?

[quaint flame]

I think it does

Easy to test

[plain trout]

Are there known issues with LAMBS_danger? We just tried to switch our modded server to profiling and lambs didn’t play along

[cloud nacelle]

Maybe it's best to ask on the LAMBS discord server, link there can be found on the workshop pages

[whole cloud]

No

[kindred radish]

Did anything change with playAction ?
I used playAction with a Move (CfgMovesMaleSdr) instead of action and it worked for years.
Now it doesn't work anymore

[whole cloud]

Afaik no

That could be related to the enum things. Can I get repro script?

[kindred radish]

Not sure how to repro since this move is from the ExileMod.
Maybe it was a bug that it worked for all the years.
https://pastebin.com/hQwj9UsQ

[whole cloud]

Well surely can do the same with some vanilla move?

[stark falcon]

I also started getting this on most recent profiling, we are running ace_killtracker (which has a mission CfgDebriefingSections entry)

[quaint flame]

It doesn't always appear for me. Haven't figured the pattern yet.

It might have to do with the screen that shows the killed entities.

[whole cloud]

If it appears, then there is an error in the rscdisplaydebriefing.sqf
format errors are obvious, the number of arguments missmatches.
Unless the format is getting a dynamic array, its a trivial fix and no need to have a repro if you can clearly see it

        _descriptionText = gettext (_endTemplate >> "description");
        if (_descriptionText == "") then {_descriptionText = "%1";};
        _descriptionParams = missionnamespace getvariable ["RscDisplayDebriefing_params",[]];
        if (typename _descriptionParams != typename []) then {_descriptionParams = [_descriptionParams];};
        _descriptionText = format ([_descriptionText] + _descriptionParams);

ayy
"CfgDebriefing" >> _end >> _endTemplate >> "description"
Assume that's in description.ext somewhere

[feral harness]

throw a ticket in my name Im doing some data

[whole cloud]

https://feedback.bistudio.com/T182822

[whole cloud]

But it doesn't set any "description"
Maybe its related to the ending type?
Mod config class CfgDebriefing has a few with description, but none of them have format arguments.

[whole cloud]

Oh..
if (_descriptionText == "") then {_descriptionText = "%1";};
Is it 0 provided 1 expected?
Then
_descriptionParams = missionnamespace getvariable ["RscDisplayDebriefing_params",[]]; should have some default value in the array.

Ah nvm you put it in the ticket. Yeah the fix will be changing the default for params

[restive pilot]

No you're wrong

You can't use moves with playAction

[whole cloud]

2.16.152017 new PROFILING branch with PERFORMANCE binaries, v15, server and client, windows 64-bit, linux server 64-bit
- Fixed: loadAbs/load and the inventory UI container fill indicator counted the contents of Uniforms/Vests/Backpacks twice - https://feedback.bistudio.com/T167469
- Fixed: Inventory UI external-container load progress bar preview was not considering contents in dragged containers - https://feedback.bistudio.com/T167469
- Fixed: lnbDeleteRow did not trigger "OnLBSelChanged" events when the selected row had to be changed

If you don't want to use the Steam branch, the files are also available for alternative download here:
https://drive.google.com/drive/folders/15p9j7C2nHUt6NoVfChX4YFuqzFXzblJh
Note: There are separate Dll files that also need to be placed into Game folder.
🥸

[kindred radish]

I used it for years, same for other people and it worked. Maybe iam wrong to use it, but it worked xD

[restive pilot]

Are you sure your moves weren't just copy pasted into actions?!

[kindred radish]

Yes, i checked old scripts i made and they 100% worked. Not running a server anymore, just realised it while testing something for someone else.
Also just tried with stable version but behaves the same, must have been a change a bit longer ago.
So doesn't really matter 🙂

[whole cloud]

Thats good news. If we broke it in the past we don't need to fix it anymore 😄

[knotty wraith]

6:32:45 Error in expression <criptionParams];};
        _descriptionText = format ([_descriptionText] + _descriptio>
16:32:45   Error position: <format ([_descriptionText] + _descriptio>
16:32:45   Error 0 установлено элементов, 1 ожидалось
16:32:45 File \A3\Ui_f\scripts\GUI\RscDisplayDebriefing.sqf..., line 45

new PROF

[analog acorn]

#perf_prof_branch message
A fix is already in place, but it requires data changes which I believe means it can't come to prof, only dev, since prof only changes the exe and uses stable data

[heavy galleon]

Linux server user here reminding the Addons/addons thingy

[autumn timber]

Seconding that one

[silk pewter]

wasn't it already sorted? (I take it wasn't, otherwise it would not be brought up ^^)

[rain moth]

only pbo names, not the addons folder itself

[heavy galleon]

Exactly. addons/IhAtElInUx.pbo works,
Addons/ihatelinix.pbo does not

[whole cloud]

😡

DayZ solves it but iterating through every file in mod folder, and doing a case insensitive name compare for "addons"

I can, I just don't want to

[silk pewter]

path.ToLower(), no?

[whole cloud]

No

That's what it already does

[heavy galleon]

But, but, but... I, and many other, would really, really appreciate it. And it really doesn't sound that hard

[whole cloud]

Its not hard, I just have irrational disdain

[worldly badge]

Stop it Lou

[silk pewter]

never! 😂

[full nova]

probably cuz linux doesn't support TFAR

[heavy galleon]

Firstly, we are talking about servers, not clients
Secondly, it does

[full nova]

no im talking about Dedmen hating on Linux

then it is

[weak panther]

I have major fps drop and micro freezes in KOTH when open 'Players' list on 2d map, it drops to 9-5
Can anyone confirm? join koth server, spawn as player, open map M, open 'Players' top-left menu
Can't get it on stable

[tribal pasture]

RPT error spam?

Players list is very heavy though, it modifies entire list each frame because engine does the same

[weak panther]

yes
https://drive.google.com/file/d/1olOvalfngg6oPw4eb8xr3til-AjjWpgc/view?usp=sharing

[tribal pasture]

format strictness

[uncut linden]

Hi. Any chance this might get attention? Or should I create a ticket?

• The inventory window bugs on setUnitLoadout - vest and backpack tabs disappear
  You can recreate this by running

[] spawn {
  sleep 1;
  player setUnitLoadout (getUnitLoadout player);
};

in the console and opening the inventory window

[feral harness]

I don't think we will do anything to it, but add the note on BIKI @silk pewter This is expected as the containers are deleted and recreated when you run setUnitLoadout.

[uncut linden]

Noted. Thanks

[topaz comet]

Does anyone also have the problem that with profiling branch you cannot connect to the server ?

https://media.discordapp.net/attachments/1155029851602948128/1261706716672823441/image.png?ex=669d29f2&is=669bd872&hm=28d8ac02d13066f9303cf4e939562d636b3de5a8f027d2d03945ccd4c1c35e47&=&format=webp&quality=lossless

[opal hound]

no

[gleaming summit]

If there is stuttering every ~5 seconds ingame, how can I best debug to find what is causing the issue?

It's likely an SQF script or at least a mod doing something that is causing it as it's only occuring when using a custom modpack

the major issue is it only seems to become very apparent when on a server with a 20+ people. Making debugging a pain due to requirement of players

[analog acorn]

You could run profiling branch (specifically prof, not perf) and use diag_captureSlowFrame to get a little bit of insight

Also make sure to note if there's any difference between different missions or different times in a mission, and if so what's present or happening when it occurs. For example, I know there are certain long-range radars that can cause regular stutters as they scan (IIRC because they scan far outside the terrain bounds and the terrain out there gets dynamically generated on every sweep)

[light cargo]

sure you're not mixing things up with either dev, mac, or something else?

[light cargo]

tried restarting steam?

[topaz comet]

This problem is showing up for more people in my group, on the profiling brench.. (Of course before I tried to write here I tried such basic things and diagnostics the only difference was always profiling brench..)

[heavy galleon]

steam is having issues, it's not arma related

It's been going "offline" randomly the entire night for me

[light cargo]

yeah uhhh, try switching steam region

its usually a temporary issue that is outside of bi/arma 3 control

[light cargo]

then its not an arma 3 issue

[feral harness]

it's steam it happens

[whole cloud]

We enabled a Steam security thing that Valve released a bit over a year ago.
But its not causing issues for most people 🤔

[topaz comet]

So actually, on the dedicated server or client side, there's not much I can do about it, but rather hope that it gets fixed (by valve) in time...?

[whole cloud]

If we get more reports of it not working, we'll just not trust it and turn it back off. We can live with less security 😄

But I am confused why so few people have issues

Is your server in some way proxied? Where its public IP isn't the IP that the players connect to?
Or does it have multiple public IP's?

Is everyone with profiling branch unable to join?
Only the non-profiling branch players can join?

[topaz comet]

People without a prof branch are just connecting just fine, but people with prof. branch for the last 3 months had various problems sometimes even when they were admins they did not have at all available on the prof. branch console which I attributed to that branch... but within 14 days now virtually everyone has been forced to stop using that branch to be able to connect at all..

[analog acorn]

Is the server itself on prof?

[topaz comet]

Not

Only clients.

[woven loom]

weird, i've connected to our community stable server just fine with perf/prof branch

are people able to play self hosted MP with your mod pack?

[light cargo]

can i ask what is it exactly?

[whole cloud]

Sure you can

[light cargo]

that question does double duty of asking to ask and literally asking

[whole cloud]

ok

[light cargo]

is arma ever going to implement steam network sockets or steam p2p just to facilitate a better user experience by using steam? nobody has the resources to work on that as usual these days?

[whole cloud]

Unknown.
Unknown.

[light cargo]

aw :(

not only lacking resources but unwillingness/laziness on your part, correct?

so, what exactly is the steam security thing that got enabled on arma appid?

[analog acorn]

I dunno if calling someone lazy to their face is a great way to get the answers you want

[light cargo]

didn't imply that

everyone makes mistakes, my apologies

[whole cloud]

I said unknown. Do you not understand what unknown means?

[light cargo]

yes sir, that does sound unfortunate, i am looking forward for it to happen in some future

[whole cloud]

@topaz comet Do you have the answers to these questions? I want to know why your server is different than others
Can you start your server weith -enableSteamLogs parameter. Then let one of those failing players try to connect, and send me the log file?

[topaz comet]

I'm sorry, but I'm waiting for answers from the provider... And I'll test the other thing tonight.

[topaz comet]

Is your server in some way proxied? Where its public IP isn't the IP that the players connect to? Or does it have multiple public IP's? - "The server is behind NAT, but outbound NAT is a different public IP than inbound"
Is everyone with profiling branch unable to join? - Yes
Only the non-profiling branch players can join? - Yes

[topaz comet]

Where can I find this log on linux server? Please.

[whole cloud]

Its just written to console output

The server is behind NAT, but outbound NAT is a different public IP than inbound
That'll probably be it. Steam requires the servers public IP to be the same as the one the clients connect to.
This is probably messing it up

[topaz comet]

In that case, I'll try to copy it into the file...

[whole cloud]

Can you not change it so that the IP Addresses are the same?
Thats a weird setup.. And Steam will likely someday enforce this

[topaz comet]

Ahh, hmm, provider doesn't have enough public IPs so some things are unified.. under one.. I'll have to live with the fact that only classic works at least that one works 😄

Thank you very much for your help in that case I guess the log is no longer needed..

[whole cloud]

The log is still needed

I'd like to know what steam actually says about it

maybe I can find a way to keep it on, but still have it work for setups like yours

[whole cloud]

You can't.
What is in profiling branch, will be in the next game update.
So leaving that, you'd be doomed.

[light cargo]

ya cgnat is very restrictive in most cases

get a vpn setup that lets you bypass the restriction

[light cargo]

not really super weird, cgnat has been the trend among isps for the last couple years

[whole cloud]

cgnat doesn't usually have two different IP addresses and them being different for incoming and outgoing?

[light cargo]

since ipv4 has ran out

it does have multiple ip addresses and everyone falls under a shared ip that you can't port forward in. p2p doesn't work

[topaz comet]

1. Try 17:14 STEAM: STC: Client: BFTDroid - SteamID:76561198109736811 - Remote ticket validation response: 0 : OK

2. Try 17:17 STEAM: STC: Client: BFTDroid - SteamID:76561198109736811 - Remote ticket validation response: 10 : Unknown ticket result
   and how upload rpt here? on phone is disabled to upload file.

[whole cloud]

You can send it via DM

[whole cloud]

But they are hosting a server, with port forward.
How is any of what you're saying relevant?

[heavy vortex]

hmm. Curious why a provider would do this:

The server is behind NAT, but outbound NAT is a different public IP than inbound

[spice dirge]

inb4 satellite connection

[heavy vortex]

lol

[spice dirge]

the only common usecase of separating inbound and outbound i can think from the top of my head

[heavy vortex]

Maybe issues with collisions between incoming ports on one machine and outgoing on another?

[topaz comet]

He doesn't have enough public ips to make them all 1:1

[heavy vortex]

I figured that was kind of a solved problem though.

Well, NAT in general is understandable.

Just not sure why the incoming/outgoing IP split.

[analog acorn]

my conspiracy theory is that many service providers (of anything, not just internet) actually find it fun to make things inconvenient for their users