perf_prof_branch

[whole cloud]

did you try the old version today? is it "works fine" or "worked fine"
It could also be a battleye update

So its only that battleye message, and battleye RCON? Sounds like a bug in battleye

[fickle spade]

the same BE version on profiling and public branch. BE v.1.218.

Switched back to public branch and problem gone

[whole cloud]

But we didn't touch that code

And why is it only noticed now, a month late

[fickle spade]

don't know about others but I've just tried to switch to prefprof branch while investigation of "BE Client not response" problem (actually not helped)))

long time before I was running public branch

[whole cloud]

Must be battleye issue, told them. We'll see

[brave barn]

Yes, https://imgur.com/a/pYE740n

[whole cloud]

Okey thank you, I think I know whats going on, not really sure why but we'll figure that out

There is only one BE change between 13 and 14, and I don't know what it does or why it was done

[empty goblet]

i can confirm the IP number order is reverved btw. on our official servers too (which use the newer profiling)

[whole cloud]

BE confirmed, fix incoming soon-ish

[fickle spade]

great. But why its behavior depends of Arma branch?

[empty goblet]

because change in the profiling branch affected it

also the server crash on missionsToServerRestart was found and fixed so next profiling i hope

[whole cloud]

3 messages above yours

This is still partially broken, it can crash on server shutdown

[empty goblet]

fixitfixitfixit

[hexed verge]

Did anything change with requirements? Updated to latest version of the profiling branch and the server is now now spitting out this

linux 32bit

./arma3server: /lib/i386-linux-gnu/libgcc_s.so.1: version `GCC_7.0.0' not found (required by ./arma3server)
./arma3server: /lib/i386-linux-gnu/libm.so.6: version `GLIBC_2.27' not found (required by ./arma3server)
./arma3server: /lib/i386-linux-gnu/libc.so.6: version `GLIBC_2.28' not found (required by ./arma3server)
``

[hushed seal]

If I had to guess 😅 @hexed verge

[hexed verge]

yeah i skipped that version as it was not too relevant on server side + had not much time that week

i reverted to stable branch in the mean time to not torpedo games from my unit
tho will this change stay and will it also be in the stable branch in the future?

[hushed seal]

By the sounds of it it's not supposed to break so probably best to let @whole cloud know that 32bit Linux seems broken.

[hexed verge]

i am fairly certain it is not broken, it is not supporting older linux distros anymore.

[hushed seal]

How old is your distro?

[hexed verge]

old, because legacy and everyone is scared to run the distro upgrade

[hushed seal]

Fair, I'd be scared to

[rain moth]

That’s how your system gets hacked 😛

[vivid rune]

My docker arma 3 server 32bit linux profiling is reporting this:
./arma3server: /serverdata/serverfiles/libstdc++.so.6: version CXXABI_1.3.8' not found (required by ./arma3server) ./arma3server: /serverdata/serverfiles/libstdc++.so.6: version GLIBCXX_3.4.20' not found (required by ./arma3server)
./arma3server: /serverdata/serverfiles/libstdc++.so.6: version `GLIBCXX_3.4.21' not found (required by ./arma3server)

Two questions:
Has the docker to be updated or the hosting? OS?
If the docker container has to be updated, will it run also the vanilla arma server with these new files?

[rain moth]

Container image needs to be updated. If new image have sufficiently new glibc then server will run

[whole cloud]

Thats why I asked who is running 32bit.
32bit now has the same minimum requirement as x64

I'm not sure if we really want to fix it.
32bit is very few users, and combination of 32bit and very old OS even fewer, and not sure if I really want to support people who are running extremely outdated systems.
This will also come to stable branch though, so you really want to update your server

[vivid rune]

for me this is no problem, if only the docker container has to be updated. And the number of 32bit users maybe will decrease even more if the " - Fixed: Game was unable to launch on CPUs from 2008 or earlier" will work for these guys. But i have to wait for the "docker creator" to update before i can to try out the 64bit linux version.

After i wrote this i thought "Why not try the 64bit version? This is not related to the 32bit files?" But I become the Seg fault as usual. So the new patch wouldn't work for me. 😦
Arma 3 Console version 2.08.149650 x64 : port 2402
11:19:17 Host identity created.
/opt/scripts/start-server.sh: line 63: 97 Segmentation fault ./arma3server_x64 ${GAME_PARAMS}

[hexed verge]

understandable, yet annoying.

[vivid rune]

@whole cloud
The creator for my docker is using debian.
The actual debian (Bullseye, released 2021-08; supported until 2024-07) is using glibc 2.31
in the next (bookworm) (testing stage) there is 2.33-7
also the sid
and the experimental is using 2.34-0
So i would say it is a little bit to early to switch to 2.34 from debian stand point.

[vivid rune]

I've done some search. The files
CXXABI_1.3.8
GLIBCXX_3.4.20
GLIBCXX_3.4.21
are part of GCC 4.9.0 and GCC 5.1.0.
See: https://gcc.gnu.org/onlinedocs/libstdc++/manual/abi.html
So I would ask:
Do I need to install now GCC for running ArmA Server?

[heavy vortex]

Well, it's GCC's runtime libraries rather than GCC...

[vivid rune]

Ah ok. I am a little confused: Is this a bug that i need this now for running the arma 3 server executable or are these "normal" needed files?

[heavy vortex]

Sounds like it's normal now.

[patent dome]

We recently updated our Linux builder, as we kept running into odd compiler errors and bugs. We should be building all 32bit and 64bit binaries for Linux on the new builder from now on, which builds using glibc 2.28. We won't be using new versions of glibc, because as of 2.31 they got rid of some -ffinite-math-only stuff needed by one of the libraries we use.

[autumn timber]

Do I need to install now GCC for running ArmA Server?
It's "Arma" (not "ArmA") and no, you don't install gcc. You should upgrade your OS to something that has glibc version 2.28 or higher. See here for the list: https://repology.org/project/glibc/versions

Alternatively, you can ask BI to be so kind to use an older Docker image (or an older OS) and install the newer compiler there and then build Arma using that setup.
This way, the newer compiler will build Arma executables using the older glibc

[autumn timber]

For the record, if anyone is interested in this, the Python devs figured out a way to build "wheels" (kinda like python's .deb files) that are compatible with a large number of linux distributions.
Details can be read here: https://peps.python.org/pep-0513/ (it's the oldest document, from 2016 and has been superseded since, but it's the one the explains stuff more in detail).

TL;DR: they were using old (but still supported!) CentOS docker images with a limited number of .so files as dependencies (as explained in the PEP above). They have since switched to using Debian as well.
The docker images are here: https://github.com/pypa/manylinux#docker-images
cc: @whole cloud (if it even matters to you?)

(however, personally, I'd love to see the 32bit version wiped from the face of the earth as it's causing nothing but trouble to me 🙄 )

[whole cloud]

./arma3server: /lib/i386-linux-gnu/libgcc_s.so.1: version `GCC_7.0.0' not found (required by ./arma3server)
But I think that lib probably shouldn't be required right?

So I'm confused by the discussion above, is there something we need to fix now, or do people just need to update their OS/Docker images?

[autumn timber]

Either people update their OS to have a glibc/xx version that is at least as new as the one on BI's build machine or BI uses an OS that is older to build

The issue here is that these distributions usually still have LTS versions that are still supported and are still getting security patches, so you can't really tell people "you're using an old, insecure OS, upgrade now!"

[vivid rune]

As i understand is the problem here is that GlibC is capped by version 2.28. So "having a newer OS than BIs build machine" would not work.

[autumn timber]

Glibc is backwards compatible so anything you build using an earlier version and be dynamically linked with a newer version

[vivid rune]

"We won't be using new versions of glibc, because as of 2.31 they got rid of some -ffinite-math-only stuff needed by one of the libraries we use."

[autumn timber]

Ok: Almost always backwards compatible 😆 EDIT: my point still stands, see below

But I'm guessing this is something that happens during compilation. AFAIR, once something is compiled, it should then work

See this, in general: https://peps.python.org/pep-0513/#key-causes-of-inter-linux-binary-incompatibility (and just ignore the "python" keyword 😛 )

[autumn timber]

https://sourceware.org/pipermail/libc-alpha/2020-May/113816.html

So you have some .o or static libraries that are compiled with the old
headers from previous versions of glibc and then trying to link with
the latest version of glibc?
That is itself is not a supported use case IIRC for backwards
compability of glibc.

So as i understand it, this only implies issues when compiling things with a newer glibc version, not the end-user not being able to actually run them

[vivid rune]

Hey, I am not so much of a linux crack. I use a docker container on a web interface.
But i have also my problems with that:
The docker creator uses an actual debian with glibc 2.31. Is this a problem from now on? Runs every linux user for arma in this issue to have a too new version of glibc?

[autumn timber]

From my understanding of the issue (and I have spent several weeks reading on it, to make sure my linux Pythia extension works on everone's machine since I'm a freak 🙄 ), and also according to Python devs, using any OS/Docker image that has Glibc >= 2.28 should fix that particular issue.

I have stumbled onto other issues, while doing this, but these issues were not glibc issues (namely Ubuntu and Fedora removing libcrypt.so.1 in newer releases, but I believe Arma executables do no link to it so this should not be a problem here)

[vivid rune]

So what have my docker creator to do? Install something from somewhere into the docker container?

[autumn timber]

@vivid rune

Pick a base distro from there that has 2.28 or higher

Looks like Debian 10 (Buster) has 2.28 so since I seem to recall you said you're using Debian, then using Debian 10+ should be good, for you (if you need a simple "use this!" kind of answer)

[vivid rune]

The docker has debian bookworm with glibc 2.31! So "higher" seems not work.

[patent dome]

If only this new one doesn't work, then there is probably some other issue. The old builder compiled the binary with glibc 2.13, which is much older.

[autumn timber]

I reread your messages above and you don't have issues with glibc but glibc++!
So maybe see this answer https://askubuntu.com/a/582910 (install/upgrade libstdc++6), although I don't have any arma linux server laying around to test this for you, so you'll probably have to google a few pages if that link doesn't solve it for you

[gritty wasp]

I don't know what happen and where but found such error in RPT.

22:56:04 Error in expression <ects [_col_abs, _wires select 0, _wires select 1]);                
};>
22:56:04   Error position: <select 1]);                
};>
22:56:04   Error Zero divisor
22:56:04 File A3\Structures_F_Exp\Infrastructure\PowerLines\Scripts\column_ruins.sqf..., line 25

worldName=hellanmaaw
Version: 2.08.149626

[silk pewter]

_wires most likely contains no items
oh, it's vanilla content 😬

[gritty wasp]

script - yes. Map - no

[silk pewter]

theeeeeen… mayhaps tell the terrain maker ^^

[boreal wigeon]

Tried to AccessTargetList for non-local AI group while non-local targeting was disabled! Engine will now enable non-local targeting which will hurt clientside performance.
Is it possible that the group id, name, whatever just any identifier, gets printed out when this message appears?
I'm trying to figure out which group my headless client is trying to access while it shouldn't.

[eager pond]

@boreal wigeon how do i disable "non-local targeting"?

[boreal wigeon]

It's disabled by default AFAIK. But you can't disable it once it has been enabled.

[cold vale]

It's when AI commands are run where the AI is remote, its an optimization that only works if the only thing that accesses sensors on AI is the local owner.

[eager pond]

😀 I need to use this branch for that?

[boreal wigeon]

Nope it's on stable.

[whole cloud]

Shouldn't be?

[boreal wigeon]

Oh then Idk. I though it got added with 2.08. As it was in 2.06 perf.

[whole cloud]

It should be a perf only change that doesn't go to stable

[heavy vortex]

If you want to force off remote target checks, there's disableRemoteSensors. I don't know if it works though.

[whole cloud]

You can't fix it

[vivid rune]

Hi,
I would ask if it makes sense to parallel the "Updating base class" things on startup of a server? I see that this is single threaded. These are all separate tasks with no dependencies, right?

[whole cloud]

config loading depends on stringtables, and config merging is sequential, order matters and they depend on whoever previously patched the same class

[eager pond]

I tried 'disableRemoteSensors true;' on server and clients (one player) with 650 AI units on server. No change in FPS 😔

[silk summit]

Most Player only "Life" servers run that command by default because they don't use Ai on their servers and there used to be a FPS increase. Weird that that is no longer the case....

[heavy vortex]

Yeah, it seemed like a sensible way to get more client perf, but it seems to be dead.

[whole cloud]

2.08.149689 new PROFILING branch with PERFORMANCE binaries, v20, server and client, windows 32/64-bit, linux server 32/64-bit

• Fixed: Non-English TreeView search was not working as expected - https://feedback.bistudio.com/T118270
• Tweaked: Logging to the Windows dedicated console log is now asynchronous - https://feedback.bistudio.com/T155902

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

[brave barn]

Reverted player ip is not fixed(

[whole cloud]

Not our fault, our update won't fix it
BattlEye already fixed it but seems their update isn't published yet

[weak panther]

Entities search in Eden not working more
https://feedback.bistudio.com/T118270#2338655

[whole cloud]

Profiling branch yes? not dev branch

"all ok in stable and 149685 DEV" ah yes. ok thanks

[merry shuttle]

Dedmen, will the RscTitles Unload EH issue be fixed before RC? Because currently in both, performance and dev branch Unload EH fires immediately after you do a cutRsc even for resources with a duration of 1e+011

[whole cloud]

RC is already in internal testing. Why do you only tell us that now?
Need repro. I don't fully understand what you mean
The unload EH is fired in the destructor of the display, so when it gets unloaded, the unload EH fires, which I'd think would be expected behaviour?

[mossy anchor]

Believed to be a issue with the falcon's model p3d, the faraway resolution lods probably doesn't have a hidden selection assigned (exocet) as the falcon after moving some distance away will appear to return to its normal camo despite if you apply a material or texture script on it.
Both variants of the Hurons when placed as compositions will return to their normal camo despite it if you swap the camo to green or black and using the script to change the material/texture.

[whole cloud]

no repro yet?

[hexed verge]

maybe a long shot. I encountered this in ace, which i fixed yesterday by removing the unload eh from the RSC

https://github.com/acemod/ACE3/pull/8969

also did a very small analysis here.
#976224710385877002 message

[whole cloud]

Yeah but that is a bug in script, caused by our broken eventhandler being fixed now.
If your eventhandler script only works when the eventhandler is broken and doesn't fire, then I don't consider that an issue on our side.
If there is an issue where the eventhandler fires when it shouldn't, then I'm interested, but when broken configs break because everything works correctly..

ACE had the same issue in Kestrel

[hexed verge]

hm, okay. weird, then i wonder where the issue in the code lies. oh well...

[whole cloud]

Well you said so in your code

Removed the "unset" on display unload as it created a race condition when a new display was opened
You have a race condition.
Unload event gets called when the thing is unloaded, and the code is written in such a way that it causes a race condition?

Replacing a cutRsc effect works exactly like you would expect (if you knew C++ land)
newEffect = CreateNewOne // Fires onLoad
oldEffect = newEffect // replaces old effect with new effect, old effect is now not used anymore and gets deleted, onUnload fires

I could change it to

oldEffect = null;
newEffect = CreateNewOne
oldEffect = newEffect

but that would require quite large changes in code

[hexed verge]

I do appreciate these answers, I hooked into the cba event for opening and closing the display to see if something is goingon there.

just one last question, does _display closeDisplay 0 trigger the unload event?

[whole cloud]

it should. unload is triggered in destructor of the display, so once it gets deleted from memory it fires

I'll change it so that it first gets cleared, and then the new one created.
This started as a quick one-liner fix to a 8 year old bug... and caused nothing but headaches till now

[hexed verge]

okay then something is fucky still.

lemme first check something before i try to do a repro

[whole cloud]

Actually, if you do closeDisplay, the cutEffects list still has a reference to it even if its closed
But I think closeDisplay should have its own handling that makes sure its unloaded?

closeDisplay just sets a exit flag.
I think for cutRsc that command might just do nothing?

[hexed verge]

not sure i just made a vanilla repro of the unload eh not firing but it fires on the next create of the display.

if you want i could send that small test mission over

[whole cloud]

if its a bug then yes

[hexed verge]

closeDisplay not triggering the unload eh seems like one to me?

[whole cloud]

is it actually closing the display?

or is closeDisplay doing nothing?

[hexed verge]

the rsc i am creating makes the screen dim, it is with one control that is the whole screen with a translucent black.

when i use closeDisplay the dim is gone and all is normal. yet the display lingers around in uiNamespaces where i have written it into for reference

[whole cloud]

so its hidden but not really unloaded, the variable reference to it also stays valid

[hexed verge]

seems like it. the idd is -1 in the RscTitles config, that stays

[whole cloud]

display unloading works by setting a "delete me" flag.
And then next frame, something comes along, sees that flag, and deletes the display.

But there is no code that iterates through the cutRsc's every frame to check if any want to be deleted.
And I don't really want to add it for performance reasons.

https://community.bistudio.com/wiki/cutFadeOut Would properly close and destroy the cut effect

[hexed verge]

i need to go on lunch break now, i believe you do as well, so i just put this here.

will try cutFadeOut in a bit

[whole cloud]

I could add a check into closeDisplay, if its a cut effect then it will do a fadeout with duration 0 instead

nevermind.. cutFadeOut doesn't terminate...
The terminate function is empty

It only works for timed title effects, which have their own command titleFadeOut (which we shouldn't really have as its a copy paste of cutFadeOut's code..)

Ah I am mistaken

But there is no code that iterates through the cutRsc's every frame to check if any want to be deleted.
There is!
But it only checks for title effects that have been fully faded out 🤣
But not for any other displays that have been exited

Because they can't
Ah this nice mess :3
The display is a child of the title effect, it child wants to be gone, but the effect itself doesn't

[hexed verge]

oh god did I give you tons of work now?

[whole cloud]

I just fixed it...
But it might be too dangerous to be merged into stable now

So currently, closeDisplay sets exit flag but the display stays alive.
The display is also still rendered every frame, but the Draw function has a check for the exit flag right at the start and skips drawing if it should be exited.

Now after change the cut effect will delete itself on next frame after its display was closed

[hexed verge]

but yeah seems like the best bet is to never unset the variable and let the engine handle it, the unload does indeed fire on a new create of the same RscTitle

[whole cloud]

First change will be merged to stable

So

load dp1
load dp2
unload dp1
unload dp2

will change to

load dp1
unload dp1
load dp2
unload dp2

That might also fix both the ACE race conditions?

[hexed verge]

probably i have not looked into the kestrel yet, but if it also unsets on unload event then it is basically the same issue

[whole cloud]

yes it does, but i don't know if onLoad would set it. I think just the unload was breaking it

[hexed verge]

from my logs from the mission i sent it is

open dp -> chat message "open"
close dp -> nothing
open dp -> chat message "open" then "close" on same frame i think --> issue

[whole cloud]

Fix probably today

[hexed verge]

i hope that also fixes the "unload event fires immediately" issue as i think that it is the same

[whole cloud]

more context?

[hexed verge]

refering to this one

[whole cloud]

As I told Xeno, need more context
onUnload fires when the display is unloaded.
"Fires immediately" doesn't make sense, there is something else going on. For example replacing a old display and the old display firing it, but NOT the new display firing "immediately".

[empty goblet]

no good deeds go unpunished

[silk summit]

Those bugs are so old for a reason (No one dared tackling them before due to the complexity and the skill required for the fix) ... still:

1. Those must be the most rewarding bugs to fix (once you are done with them)

2. You are basically the only person in the world with the skills and position to do something about them (kk too 😅 )

3. Your fixes are enjoyed by tens of thousands of players, in contrast to the Reforger fixes which are currently enjoyed by 200 players

4. Considering the state of Enfusion we are still several years away from a proper Arma 3 replacement (Unless BI pulls a CD Project Red on us with Arma 4). So Arma 3 revenue will still help fund the salary of many devs for years to come and so any huge time investment you do to the game ensures that your peers at the office keep their jobs, specially in these obscure economic days that Europe has ahead

[fickle geyser]

Ouch.

[silk summit]

This channel is exclusive to Arma 3 devs right? 😅 Don't want any Reforger dev reading this... (well I am just being honest)

[weak panther]

I can't make a PhysX object as simple until I disable simulation for it (in Eden attributes)
Is this changed intentional or a bug?
↑↑in Dev and Prof branches↑↑

cannot repro in Stable

[weak panther]

see on YT https://www.youtube.com/watch?v=wNHp14BSQYo

[woven loom]

Hey, been a long time since i've been around here. I was wondering if there's any updates on the apparent memory leak while using 4K textures? It's been an issue ever since Global Mobilization released, making very high or ultra textures unusable for me. Made a post on the forums in March with a list of relevant tickets i've made over the years but there's no update on any of them so far. https://forums.bohemia.net/forums/topic/160288-arma-3-stable-server-208-performance-binary-feedback/?do=findComment&comment=3455538

It's not great having to play GM with visibly lower quality textures just to avoid CTDs. Earlier the problem appeared to be RAM/page file related, now it appears to be VRAM related. My system hasn't really changed save for windows 10 and driver updates, and of course overkill on pagefile (i'm running 16GB pagefile).

Not had enough time in life to test with other mallocs etc.

[silk summit]

I got FPS/stutter issues when enabling both GM and SOG simultaneously. Not crashes but increased stutters and lower FPS that I never had with any other mod combination.

I don't know if SOG also uses 4k textures but honestly I think 2K would be just fine. Perhaps GM can lets us choose 2K texture installations

[woven loom]

I think selecting lower texture settings should do that, but i basically get the same issues in very high and ultra. so i have use high which doesn't look that good close up, and when mixed with vanilla assets it's a bigger problem because they just look straight up bad

Also the step up in VRAM usage between high and very high is pretty big

[whole cloud]

No, the "Local Only" option should be hidden and only appear after its disabled or simple

[whole cloud]

no

[weak panther]

Iam about "Simple Object" option, its hidden if simulation is enabled

[whole cloud]

Yeah seen that, I'm testing now

2.08.149712 new PROFILING branch with PERFORMANCE binaries, v21, server and client, windows 32/64-bit, linux server 32/64-bit

• Fixed: nearestObjects classname check didn't work on Ammo - https://feedback.bistudio.com/T166472
• Fixed: Entity search in 3DEN was broken since last update - https://feedback.bistudio.com/T118270
• Fixed: cutRsc would first call onLoad and then onUnload handler if a existing display was replaced (order is now Unload/Load)
• Fixed: closeDisplay on a cutRsc effect, would not remove the effect

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

[full nova]

wait new stuff??

[whole cloud]

Ah found the simple object bug
Wrong line in a switch/case, it was apparently supposed to fall through..
Thank you!

[mossy anchor]

Bump message for the falcon drone issue

[unborn urchin]

profiling branch is immediately crashing for me when running vanilla

boots into the menu fine, can go into 3den, but when I place down a unit on VR terrain and hit play in singleplayer, gets stuck on the loading screen and crashes

[whole cloud]

Hahahahah fuck..
we pushed 149712... 149713 was a crash fix for a thing I forgot to check

[unborn urchin]

ah so you won't need the crash dump then? 😅

[whole cloud]

correct

ETA 2.5 hours or so

[uncut sigil]

https://tenor.com/view/the-rock-rock-gif-21708339

[night mulch]

https://tenor.com/view/rock-one-eyebrow-raised-rock-staring-the-rock-gif-22113367

[eager pond]

This server command started to crash from some days ago until now, on profilling/performance branch. I reverted to stable and it's ok now.

"password" serverCommand "#restartserver"```

[silk pewter]

ha ha - fail

[night mulch]

So this is what i get after everything i’ve done for this server

[silk pewter]

it's so rare you fail I have to make it count ❤️

[whole cloud]

Can you send me crash dumps for that? .mdmp file

[night mulch]

Then you better keep counting 😈

[eager pond]

I will try to make a repro

[whole cloud]

i don't need a repro, I just want the crash dump

If that shows me where it crashes and I can see whats wrong, thats enough

[lofty vigil]

After the v21 profiling update my game crashes immediately after I finish downloading the mission file of the server I play on. Switching back to the stable version of Arma 3 solves the issue I have the mdmp file for you guys to look at if that would help you debug this issue feel free to add me and send me a dm for the file

[eager pond]

@lofty vigil this is a know problem: #perf_prof_branch message

@whole cloud sorry to bug you. No crash anymore. Can't generate DUMP file.

[lofty vigil]

gotcha just wanted to offer any help I could with mdmp file or anything if that helped them

i really should learn to read this discord more because I spent like 20min trying to figure out why I was crashing

[woven loom]

So is this effectively a "won't fix" or just "not priority"? Both fine, just asking so I can let it go lol

[whole cloud]

A "I wouldn't know where to start looking and don't want to invest potentially days on this"

[woven loom]

alright, thanks!

[full nova]

is the new profiling dropped yet?

[full nova]

or in ^

i was confused

[patent dome]

Some issues with the linux builds, ETA

Will provide an ETA once I have solved the issues.

[hexed verge]

(discord did not scroll down, disregard, crash issue already reported)

[patent dome]

Updated ETA: 2h, so around 17:00 CEST

[whole cloud]

For those who need it early, perf x64 v22 https://s.arma3.io/arma3_x64.exe

[silk pewter]

pssst, hey kid - like profiling branch? here, look at this! ↑

[uncut sigil]

Thanks yall

[inland dew]

with 712 Arma crashes (STATUS VIOLATION) all the time for me, when trying to join any multiplayer server
had to go back to stable to be able to play

[merry shuttle]

No, no repro yet. No time yet. Am ill.

[patent dome]

Known, fix isn't too far out

[spiral pond]

if you switch to another branch (non perf/prof), overwrite the exe manually with perf/prof - will updates to the other branch overwrite the exe again? or only stable branch will?

(in technical terms are branches always all data, or can a branch only affect specific parts - ie exe and bin folder - and leave the rest unaffected)

[whole cloud]

update will overwrite again

[spiral pond]

ok ty. so its recommended to use to launch the "custom exe" from modfolder essentially? (or use symlinks to combine both)

[whole cloud]

just replace the main exe with the new one, thats the same that steam will do once it has the update

[whole cloud]

2.08.149723 new PROFILING branch with PERFORMANCE binaries, v22, server and client, windows 32/64-bit, linux server 32/64-bit

• Fixed: Eden SimpleObject object attribute option was only shown if a object had simulation disabled
• Fixed: Crash from v21

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

[whole cloud]

[floral socket]

is it up on steam?

[whole cloud]

should be

[spiral pond]

from what i understand, i would need to do manually replace the exe (and dta folder) each time the other branch gets updated tho

[whole cloud]

gets toppled over

Dahlgren would probably know that more than I do

[spiral pond]

steamCMD probably could automate the process?

[fickle geyser]

steam depot downloader or steamCLI should allow you to download only the depot with the exe, if there's separate one.

[rain moth]

what end result do you want to achieve? 😄

[spiral pond]

mix perf_prof_branch with creatordlc

[rain moth]

ok, I would do like veteran suggested and download profiling (and performance) depot to a separate folder. you can symlink to main arma 3 folder with different names as needed.

depot id is 249503

let me know if you want specific command, I think I have one for our windows arma 3 server

[spiral pond]

would be handy for steamCLI as its not immediately obvious what parameter combo are needed

[fickle geyser]

steamcli --username YourUser --password YourPassword --app-download --appid 107410 --depotid 249503 --targetdir ./arma --synctarget --branch profiling --branchpassword CautionSpecialProfilingAndTestingBranchArma3

this downloads/updates following files in the specified target dir:

    arma3.exe
    arma3profiling.exe
    arma3profiling_x64.exe
    arma3server.exe
    arma3serverprofiling.exe
    arma3serverprofiling_x64.exe
    arma3server_x64.exe
    arma3_x64.exe
    steamclient.dll
    steamclient64.dll
    tier0_s.dll
    tier0_s64.dll
    vstdlib_s.dll
    vstdlib_s64.dll

[rain moth]

steamcmd sadly requires the manifest and can't figure it out on its own, depot downloader and steamcli can

[empty goblet]

ye i use separate folder on the prof branch download too, just auto copy the binaries (in rare cases libraries or data but that's manual touch)

[rain moth]

DepotDownloader.exe -app 107410 -depot 249503 -beta profiling -betapassword CautionSpecialProfilingAndTestingBranchArma3 -username %STEAM_USERNAME% -password %STEAM_PASSWORD% -dir %OUTPUT_DIR%

[eager pond]

Dedmen, i have the DUMP for servercommand #serverrestart crash.

The ziped file have 7 megas.

[whole cloud]

[eager pond]

https://drive.google.com/file/d/1CVQRW-vhM6C7uaFIhf_PfKvdoy6jtp6S/view?usp=sharing

[whole cloud]

Something is very wrong with that.
Thats not running #perf_prof_branch, and also the Arma exe thats running there doesn't match the one I have?

[eager pond]

Hmm.

May be if i use steamCMD?

I will try. But need to wait for another restart.

😕

[eager pond]

Used SteamCMD perf branch to update the server:
It crashed 3 hours and 40 minutes ago: https://drive.google.com/file/d/1EJ8wfYwh2vsb1RCzIKcV9lF6Ow2916yz/view?usp=sharing

[vale shoal]

Actually if I remember correctly, this crash is a very old bug. Happend for me years ago when we used the restart function via infistar.

[eager pond]

Oh! I can live without "#serverrestart" 👍

[whole cloud]

Thanks... interesting...
Game was shutting down, cleaning up all vehicles.
But some helicopter thought the correct thing to do while game is shutting down and is trying to delete it, that now would be a great time to initialize itself and get loaded while everything else was busy unloading 😄
Nice find

[silk pewter]

helicopter should crash itself, not the game!

[eager pond]

It's a CUP mod Heli?

I use those mods: Immersion Cigs / Swin Faster / Advanced Urban Rappelling / CUP Vehicles, Weapons, Units 1.17.1 / CBA v3.15.7

[eager pond]

Heli is a rebel!

[eager pond]

I'm trying this before "#restartserver":sqf {deleteVehicle _x;} forEach allMissionObjects "";Edit: I was smart enough to change clock time and simulate restart 🥲 ... and it's fixed now!

[grizzled gate]

Is the profiling version fixed?

[worldly badge]

Yes already