multiplayer

[woeful trout]

@meager spade

@meager spade

this is the code

[worldly needle]

Take a look at MVC pattern and understand it in wide spectrum. It will give you idea on how

[woeful trout]

so the code is wrong?

i mean the logic behind the replication

or im missing something

[worldly needle]

yes your code is wrong. You are expecting to change replicated value from UI.

[woeful trout]

i found another way to change visibility when it picks up a weapon from the weapon

[worldly needle]

and even you can achieve that “that replication is non necessary”

[woeful trout]

this bp is my actor (item pick up)

it works well on server but no on client so if i fix the replication maybe it will works

what do you think?

@worldly needle

[topaz knot]

I'm still trying to learn multiplayer. So I've been doing server to client for everything, but it looks like mobs like chest or something that spawn mobs shouldn't go server to client? Multiplayer is a pain. Could someone nicely explain this to me.

This is a Cabinet that on game start picks some loot and an enemy to possible spawn when the play opens the draws.

[tame sapphire]

Is there a better way of adding a widget to all player screens via the server than using the game state and a foreach loop on the player array and either calling a spawn widget event on player controller? or player HUD? or would a multicast in the game state be better? (Ideally id like this to delay and hold up the turn for the length of time the widget is on screen?)

think turnbased and a widget to say starting server turn and another for player

[dark edge]

spawn a replicated actor on the server

If you're learning, just delete all multicasts from your project. Use replicated variables and RepNotify.

[worldly needle]

sorry, I had take off. Look your function names and variable names are very vogue. I dont understand single thing to help you.

This is not going to help you either on long term project.
You need to give proper names to your functions and variables first which you can understand even you open this project 5 years later. And that will help you a lot even finding a help.

Right now I'm seeing bli, serv kinda function names, new var as variable name which is being replicated. What they are being used for. How you dont forget what these things are doing ? There are many question marks in your implementation and they are piling up on each.

But dont worry, you can fix and put them into perspective with doing some reading and understanding logic of producing, rather than replication. I guess you need to spend time on those.

[woeful trout]

new var is the widget ref

the bli and serv is the custom event that i use to replicate the ssystem

[worldly needle]

you missing point. I know you have all the answers about functions 🙂

[woeful trout]

the bli is on multicast and the serv is on server

[woeful trout]

ok i get it ill strugle alone and ill send you tomorow

[worldly needle]

just refactor with variables and functions and send them to me once again.

[woeful trout]

the image 43 is the image that i want to be hidden

i take with my client and the resault is to make hidden the image on server

when i pick an item on server it works on server perfectly

[nova wasp]

don't replicate a widget, replicate the information in it and if it's what you want visible

replicating an instance of a widget is not what you want to do here

you want to send information that each client can represent in the ui after receiving

also don't use RPCs for this (imo), onreps will be easier to work with every time as they are persistent won't break late joiners or require a chain of reliables

[woeful trout]

So what you suggest me to do?

@nova wasp

To solve this problem

[nova wasp]

I think the simple thing would be a variable on the gamestate that represents the current state you want to show

then in that state's onrep create the widget LOCALLY

[woeful trout]

So it will not be replicated on the details?

[nova wasp]

no, don't replicate a random object that doesn't have replicated state

[woeful trout]

Got it

[nova wasp]

the goal of replication is to send information, a UI widget is about representing information about the game in a nice way

so the information to send should be separate at least in the replication sense

[woeful trout]

Ok so in the game state I make a variable, I make the variable on rep and inside the on rep var (of I guess the widget var) I creat the logic behind the visibility?

I'm a little bit confused 😅

[nova wasp]

sure, that should work

it could even be a replicated struct or multiple variables, whatever

the on rep variable is not a widget reference here, it should be something like a bool or name or something that DESCRIBES the state you want to show

it could even be the class of the widget I guess if you wanted as asset paths are stable over the network

[woeful trout]

Ok and after I creat this var and I make the logic inside what have to do?

[nova wasp]

I'm not a youtube tutorial... you might want to check out the pins for starter resources

[woeful trout]

I mean some where must put this variable

[nova wasp]

ideally the onrep creates the widget and then the widget works as before, but represents state from this replicated thingy

the gamestate would be a decent choice here because it's replicated to all clients

[woeful trout]

True but if you have something on your head ( I mean an idea) on how to solve it I would appreciate that

[woeful trout]

I'll try that

[nova wasp]

where it should be depends on who it is sent to and why

[raw trail]

Has anyone worked with experimental MultiServerReplication plugin? This is a magic pill for MMO games?

[quasi tide]

If you can't read the source, don't even bother. It is extremely experimental. But it is what they plan on using to achieve their goals for UE6.

[sullen sail]

Is there a way to determine this directly from the engine source code?
For example, does BeginPlay run on the server only, or also on clients with simulated/autonomous roles?
I want to understand which built-in events are triggered depending on the network role of the actor.

Any tips on how to trace this in C++ or know where in the source this is defined?

[quasi tide]

Then you need to understand how networking works

All the lifecycle events will run on all machines

[nova wasp]

add {,,UnrealEditor-Engine.dll}::GPlayInEditorContextString to your debugger watches

this is nice because it lets you see if you are in a server or client when hitting a breakpoint in the editor

[lament flax]

or standalone, etc

[river bobcat]

Should I playandwait montages in GAS ability, on server side and relay on notifies? I noticed use multithreading must be disabled on AnimInstance to net getting notifies twice. Do you have any good rules or examples, documentation how to handled that in a good way? (Dedicated server)

[lament flax]

notifies is not safe ofr multiple reasons

for important stuff that needs save serve rauth, use timers

(read ntofiy time for given AM, and play timer)

https://notes.hzfishy.fr/Unreal-Engine/Animation/Anim-notifies

[meager aurora]

How would you handle spawning actor ( an arrow, while aiming) in multi action rpg. Would you spawn a client only copy for responsivenes, or just rely on server authority?

I am, inside my ability, spawning one copy for client and one copy for server, which is invisible for the owner. Wondering if it's correct apprach, or is there better?

[exotic wasp]

client prediction is nice

[nocturne quail]

Is anyone see a bug in this code?

UItem* UInventoryComponent::AddItem(class UItem* Item, const int32 Quantity)
{
    if (GetOwner() && GetOwner()->HasAuthority())
    {
        UItem* NewItem = NewObject<UItem>(GetOwner(), Item->GetClass());
        NewItem->World = GetWorld();
        NewItem->SetQuantity(Quantity);
        NewItem->OwningInventory = this;
        NewItem->AddedToInventory(this);
        Items.Add(NewItem);
        NewItem->MarkDirtyForReplication();
        OnItemAdded.Broadcast(NewItem);
        OnRep_Items();

        return NewItem;
    }

    return nullptr;
}

my doubt is i need to first NewItem->MarkDirtyForReplication(); and after add to items

[exotic wasp]

why do you need to mark it dirty?

[grizzled stirrup]

Just curious how do games with intricate enemy attacks (some boss that slams multiple locations on a specific animation rhythm) deals with this?

Surely they aren't doing magic number timer offsets for these?

I've found notifies to be very consistent but PvE game so doesn't need to be as strict

[nocturne quail]

to replicate the object, its not an actor

[lament flax]

Designers use notifies, dev read them and programmatically create timers

Nothing to fancy or hard to setup

[grizzled stirrup]

Read them how? If you just mean hooking to the notifies to fire a C++ function then that's what I'm doing

Or you mean they are basically calling timers to fire off when each anim notify would have been called from the anim playing in case the notifies don't reliably trigger?

Seems complicated if you have say a big animation with lots of different damage points like an elden ring boss with 6 hands all slamming in different places at different times

[lament flax]

You can read notifies from a AM ptr

AM -> get notifies -> get time -> bind some lambda/member func with notify class/ID

Read my link

[grizzled stirrup]

However branching point notifies don’t fire if there are 2 on the same frame, especialy common when its the start of a montage ah this explains why I haven't ran into it

I tend to have notifies spaced out by at least a bit of time before and after

and haven't noticed one being dropped ever

But also on listen server, may be required on dedi server (not sure if anims run there)

[exotic wasp]

just use replicated subobject list

[nocturne quail]

https://github.com/Yalenka/SurvivalGame/blob/main/Source/SurvivalGame/Items/Item.cpp#L90

[quasi tide]

Do note - branching point notifies and regular notifies are not the same thing.

[grizzled stirrup]

I've yet to notice a regular notify not firing unless very close to another or on the first frame of an AM

but may be other cases I'm missing

Maybe low frame rate or large amounts of notifies in a short time frame etc.

[lament flax]

I trust Vaei, who worked on big productions

Also, in a MP contexte it makes sense to use the time instead of the notifies

[grizzled stirrup]

Depends on the context, for PvE games it can be nice to have enemy attacks deal damage on every machine locally so that it feels fair for all players without them seeing the enemy wind up but on the server it has fired and they just take damage out of nowhere

But yes for PvP sounds like the server timer is the way to go for consistency

[grizzled stirrup]

I just always like to see the actual issue happening in my project before doing workarounds, I think it's a healthy way to work

[quasi tide]

I agree to an extent. There are some things that are kind of obvious to avoid based on the issues. Like CACs 😅

[grizzled stirrup]

CAC?

[quasi tide]

Child Actor Component

[grizzled stirrup]

Ah yes haven't even tried to implement something with those after hearing all the problems 😄

[lament flax]

Doesnt change the logic, just who execute it

[grizzled stirrup]

right, well I guess I haven't come across a situation yet where I've missed a notify so it seems to work at least for most cases in practice

I'd find doing timers to add a good bit of complexity as you'd need to not only know when an event is due to happen but also where so you'd have to start doing delegates with socket names passed and such

[lament flax]

This issue aside, handling all kind of notifies in the ABP seems terrible

For versioning and organization

[grizzled stirrup]

To me it is very simple, animation plays and notifies happen to tell code to do things at certain points. I don't really see how moving to a complicated timer setup makes life any easier there unless you are doing very simple animations that only need to fire off a few simple notifies

Of coures assuming you aren't running into edge cases

[lament flax]

Timers are completly abstract

Allows you to do anything

Doesnt change how you place your notifies

[grizzled stirrup]

But say for a monster that has 6 hands that each fire off a notify is it not a lot of work to set up 6 timers when the animation starts that each pass a socket name to deal damage at a certain location and also you need to keep track of things like players slowing enemy animations with freeze or such?

notifies are great as they respect retiming

[lament flax]

It makes things a bit harder on the dev side once

[grizzled stirrup]

I'm not even sure how things don't get wildly out of sync when you have things that can retime or otherwise influence animations

[lament flax]

See vaei plugin

[grizzled stirrup]

You also then have to account for animations being cancelled and cancel the subset of currently running timers

[lament flax]

As an example

[lament flax]

Nothing complex

[lament flax]

Setup would be automated with abstract methods

Not harder than comparing FNames in ABP

[grizzled stirrup]

I guess the way I see it, notifies fire off 100% of the time for me in code as long as I am not putting them on the first frame or on top of one another so it wouldn't make sense to move to a more complex system. Looking at the plugin now just out of curiosity on how it handles retiming

like an animation slowing or speeding up after the montage started

I'm not seeing retime support on the plugin, seems to assume the anim will play at a certain speed for the duration

[lament flax]

Its easy to have a notify played at the same time when multiple anims are played

Race conditions

[grizzled stirrup]

No like say an anim is played at play rate of 1, then the player inflicts freeze on the enemy which ramps down the play rate depending on the strength of the freeze

Suddenly all the timers are way out of sync from the anim

[bronze mauve]

I would think as long as the anim notify logic checks for authority before applying damage it should be fine to have as a notify right

[grizzled stirrup]

Notifies work great for me aside from the known edge cases which I simply avoid (multiple on top of one another and very first frame notifies (even those work fine unless you are constantly restarting the same montage multiple times in a very short timeframe))

Otherwise they appear to be 100% consistent (I haven't had any issues where they don't fire off but my game isn't using dedi servers so maybe things are different there if the server doesn't actually tick the anims, not sure)

[halcyon ore]

I've never had issues with notifies either, so this has me curious. 😛

[digital herald]

The server doesn't tick them by default so if they depend on socket locations they won't update unless you change that

You can make the server tick anim skeletons though. It's just off as an optimization

[quasi tide]

But that is only an issue with dedicated

[grizzled stirrup]

Yeah I am more curious than anything because I wouldn't want to do a workaround with timers when the notifies are much more convenient (free location and timed events which respects anim retimes and slowdowns)

[digital herald]

yea

[quasi tide]

Listen servers will obviously tick them

I say keep doing what you're doing Kyle and if it does turn out to be a problem, you have a potential solution.

[grizzled stirrup]

Sounds like a plan 🙂

[halcyon ore]

Wouldn't meshes always need to be ticked on server, for like player collision and stuff? (I'd hope all the collision isn't just client side?)

[digital herald]

You'd need to enable it if your game depends on that

[thin stratus]

You can, but in theory you'd only need the Capsule

[digital herald]

For the physics asset etc.

Again only for a dedi

[halcyon ore]

Oh, so if your game depends on it, then it doesn't really matter. ok

[thin stratus]

You can still be pretty strict on when it ticks. We have disabled tick on Dedi entirely for the MeshComp and the AnimInstance. Not even enabling it for RootMotionMontages. Also changed the ShouldTickPose to not tick the Pose if TickOnlyMontages is enabled and we aren't playing montages.

Does bring its own lovely set of problems, but well, performance > all

[quasi tide]

I'll just use the Engineer's motto to sum all this up, "it depends"

[digital herald]

Not even for root motion montages? Do you just predictively play them locally and multicast from the server or use GAS ability task

[thin stratus]

Neither. I manually extract the transform. There are built-in functions for that already. Just need to keep the play position in sync

[digital herald]

Interesting.

[quasi tide]

That sounds like work

I don't like work

[digital herald]

Perf gains tho

[quasi tide]

Just up the hardware requirements. Profit.

[nocturne quail]

using low quality animations don't need all these extra work stuff I think, just use AM thats all if your project depends on it

I have a struct with durations Start, End and points for all notifies... so just calling any function in code which is expected to be called by the notify in anim montage.

server using this struct, and client using the AM

[thin stratus]

Yop. We can't do that fancy stuff. Too many players. Gotta keep the server alive.

[mystic estuary]

Hello, I have some pre-placed actors with cosmetic components that run some logic to update their state. I'd like not to have that on dedicated servers, and I was thinking about not having them on dedis at all. How would you go about it? I thought about either wrapping initialization code in #if WITH_CLIENT_CODE or destroying the component on begin play, but I'm currently looking around for possible solutions

[modest crater]

Depends on how you intend to package the game as a whole, simplest approach would be a GetNetMode on Beginplay/PostInitializeComponents and check for DedicatedServer

[digital herald]

Do the actors need to exist on the server?

If so then just what Baffled said

[halcyon ore]

Seems pretty stupid, but wanna see. 😛

Is there any form of networking (that be a UPROPERTY, or some UFUNCTION)
That has the same outcome of reliable (it will replicate), but without the super forced make it happen 24/ 7 nature of reliable?

basically, I want something to replicate, but I don't mind if it could take 1 second, 40 seconds, etc.

[digital herald]

Lower the net update frequency I guess

[halcyon ore]

That affects the entire actor?
But, also it would still not mean 100% replication?
Unless, lower it, and use reliable? (But, don't they run on different things?)

[digital herald]

If it's a replicated variable or a reliable RPC it will replicate.

You can also toggle on/off a variables replication with a replication condition of COND_Custom.

[halcyon ore]

Oh, variables will replicate?
I always assumed they could get lost in packets, just like unreliable functions

[digital herald]

They will replicate.

They aren't guaranteed to be as fast as possible but they will replicate - so basically that's what you're asking for.

[halcyon ore]

Oh, ok.
Good to know.
I didn't realize variables were a 100% replication thing.
I thought that was just like the existence of an actor, and reliable functions that was 100%

[digital herald]

Replicated variables specifically.

If they aren't marked for replication, they don't replicate.

[halcyon ore]

Yes, of course. 😛

[digital herald]

But yeah you can rely on them to always replicate, unlike an unreliable RPC

[halcyon ore]

Since your here.
Does that mean COND_InitialOnly means it WILL replicate on initial actor replication.
Or, just that it will only replicate once?
Since, UE docs says it will attempt
This property will only attempt to send on the initial bunch

[digital herald]

Yes as the actor first replicates. If that replicated variables changes again after that on the server, it will not replicate

[amber stone]

Anybody here got any ideas on this? #chaos-physics message

[mystic estuary]

Check the net mode and then what? I thought to either not run any initialization code or to destroy components altogether

[modest crater]

Up to you, if you wanted to, make the components created on PostInitializeComponents then it only gets created on non dedicated to begin with. Components that do nothing but exist aren't really an overhead concern though so it really just depends on what you want to do

[mystic estuary]

Is there people using Redpoint EOS + Common Sessions? What is your experience? Does common sessions ease the workflow or not?

[steel anchor]

Hey guys!
I’m having a problem with Blueprint Interfaces and widgets in multiplayer.

Basically, when my character takes damage, I want a widget to appear on screen (like a damage indicator or something). The issue is: the widget shows up for everyone, not just for the player who took the damage.

I’m not sure how to make the Blueprint Interface call only affect the specific client who got hit.

Thanks in advance for any help!

[latent heart]

Send a message directly to teh player that got hit.

[dark parcel]

Damage should be a server execution only, if that's not the case then you got a problem.
If damage is indeed a server executed portion, then all you gotta do is make the server tell the target client to run a function that show the widget.

[steel anchor]

Here’s my code, which currently doesn’t work as intended. Thanks for the answers, but I’m still not really sure what exactly I need to change. So I’m sharing my code for a bit more clarity 😅

[crisp shard]

when you have a mesh in a blueprint, can you trigger one of the component overlap events in that blueprint (OnComponentBeginOverlap, or OnComponentHit) by overlapping it's trace response channel?

[nocturne quail]

possible to modify the state machine so it can have multi entry points based on some condition?

the issue is when you transit from one state machine to another, and if in the previous sm you were in the crouch mode and in a new state machine you start from the default entry point stand and will end in the crouch state

to make it short:
1: you are in unarmed state machine in crouch pose
2: you equip weapon and you transit to the armed state machine in crouch pose
issue: player will stand up and then go back to crouch since state machine default entry point in stand

or the dirty way will work fine: all states armed/unarmed in one state machine...

this is what i want it to work, so if previous unarmed state was crouch, the new armed entry point should be crouch...

┌───────────────────────────────┐
│ Combined State Machine        │
├───────────────┬───────────────┤
│ Unarmed        │ Armed         │
│ ╔═══════════╗ │ ╔═══════════╗ │
│ ║ Stand     ║ │ ║ Stand     ║ │
│ ╚═══════════╝ │ ╚═══════════╝ │
│ ╔═══════════╗ │ ╔═══════════╗ │
│ ║ Crouch    ║←┼→║ Crouch    ║ │
│ ╚═══════════╝ │ ╚═══════════╝ │
└───────────────┴───────────────┘

[lost inlet]

Photos of a screen are certainly a choice over a screenshot, but in your first image the "test object" pin isn't even hooked up to anything

[nocturne quail]

well seems like i have to make multi SM's for crouch, stand, prone.... and crouch sm will have default entry point crouch.

void AActor::transitionToArmed(ECharacterPose previousPose ) {
    if (previousPose === ECharacterPose::CS_Crouch) {
        ActiveAnimInstance->enterArmedCrouch();
    } else {
        ActiveAnimInstance->enterArmedStand();
    }
}

[steel anchor]

Yeah i know just don't have discord in my computer ( I corrected and connected the node right after thank you anyway )

[lost inlet]

If you go to https://discord.com/channels/@me it's a web client

[lyric heart]

What do you think about using the web and php as a service ?

https://streamable.com/4kneu0

for a simple game like this i think it's fine

[latent heart]

Nothing wrong with web/php.

[lyric heart]

it's just a lot easier i already know how to use the web

this multiplayer stuff is just so difficult lol

[lyric heart]

thats what i figured

[latent heart]

When I say there's nothing wrong with web/php, there's definitely reasons to use UE's multiplayer stuff. Right tool for the job.

[lyric heart]

ya for this little game i think if i can host a free php and just use the web stuff its pretty nice

it's long polling mostly so it's not very data intense

and i don't need live data of per character transforms

[latent heart]

Then you're probably fine.

[lyric heart]

it's all prototype so it's all a bit wonky, but it's very efficient and doesn't slow the game down at all

[mystic estuary]

How do you test dedicated servers if you want to verify the full connection from main menu to the game map? Do I have to build for that both the server and client?

[meager aurora]

Hi, I noticed that in my multiplayer setup, play as client, sometimes when starting simulation the camera is stuck inside the player and I am unable to move, sometime this happens for one player, sometimes for both.

https://streamable.com/dk6wu3

[mystic estuary]

It should be related to how you manage your initialiazation

[halcyon ore]

Networking check. 😛
Do actors/ components use any traffic, if they legit never change? (No variable change, no movement, etc)

[fossil spoke]

If nothing changes, nothing is sent.

[halcyon ore]

Ok.
I wasn't sure if UE like re-sends info, thinking there could be an issue periodically (if a replicated thing doesn't change for some amount of time)

[exotic wasp]

I'm pretty sure they add a little computational need because networked objects need all their properties checked every tick

[bitter pine]

Does DestroyComponent not replicate to clients?

[nova wasp]

the cost would mostly be polling for changes on the server's net update

bandwidth wise I don't think it will add anything worth worrying about after their state settles?

[thin stratus]

No they don't. They have different poll rates and are also mostly push based by now

[exotic wasp]

i was unaware of that

[modest crater]

Net ticks are at a fixed interval different to the regular games tick as well as actors themselves have their own net update rate and other various factors - push based is still opt in and requires you to be explicit if a variable is push based

[halcyon ore]

Does they mean by default, it does take network just being idle (since you said push is opt in)
Or, do I have my idea backwards

[halcyon ore]

For context, if it helps.
I have a component that I have in a ton of actors.
I originally had the component all controlled by RPC networking.
But, a new idea would be too annoying with RPC, so now the component needs to have a constantly replicated variable on it.
On a ton of actors.

[bronze mauve]

Is it not possible to have a replicated UObject in Blueprint?
I know in C++ you can add a replicated subobject to the actor's list, but I don't think there's a way to do it in BP. You can mark a variable as replicated, and construct an object at runtime, but it doesn't seem to actually support replicated variables on it.

You can mark variables for replication on UObject Blueprints but they don't appear to actually replicate

The UObject itself can replicate apparently, but it cannot have replicated variables on it - they don't replicate when changed serverside

Blueprint Actor Components can have replicated variables no problem though. So I assume this just isn't supported in Blueprint for plain UObjects.

[bronze mauve]

If anyone with more experience could confirm that I'd appreciate it.

[nova wasp]

actor components are replicated subobjects already

[bronze mauve]

Yes, that's what I'm seeing, and in Blueprint I assume that's why variables on my BP-only actor components replicate when they change.

But Blueprints based on the Object class don't appear to have a way to be added to an actor's replicated subobject list in BP

In C++, you need to manually do so with AddReplicatedSubobject but there's no corresponding function in BP that I can see.

[nova wasp]

it appears that you are correct in thinking that bp can't replicated subobjects like this

that said... just slapping on actor components would probably be decent enough

blueprint multiplayer is super simplified in general so I'm not surprised it can't do this out of the box

[digital herald]

Yes, it can appear confusing since you're able to mark variables as replicated in a Blueprint based on UObject - this won't work if you've constructed the object in BP - as you mentioned, there isn't a way to add to the replicated subobject list unless you're working in C++.

But yeah, if you want replicated variables on a subobject and you're working purely in BP - components are fine.

If you really need a UObject BP with replicated properties, it's time to start using some C++.

[crisp shard]

i just realized i can do a switch on a multicast for vfx/sfx and im considering doing this for a lot of these simple multicast events in the future but am i wrong to think that this is doing anything postive? it makes sense to me that i should not be bothering to spawn those effects on the server as well and the mutlicast does it on both, and i realized this by actually doing it in an onrep (BP one, so fires on server and client), and i had a switch to make an error go away and i realized i could do that on a multicast, and yea just thinking this through made me think i discoverd a thing to do for these types of events but wanted to bring it up and see if there's some flawed thinking in my approach

[shrewd ginkgo]

https://blueprintue.com/blueprint/06aq24mf/
https://blueprintue.com/blueprint/ac3qzv7n/

steam overlay comes, I set up the server but when I search, the session result lenght is 0, it cannot find the server

[grand kestrel]

@thin stratus @pallid mesa I'm gearing up to 'officially' release Predicted Movement 2.0 which has BP support including a single-cmc branch, movement modifiers, and partial client authority system
I could use your opinions tho, should single-cmc become main, this is a complete CMC/Character combo that has a typical setup, with sprint+stamina+a bunch of modifiers+client auth already setup, common to what most games would actually use, and is important for the BP support
I could make what is main now into it's own branch, but I think maybe having this single-cmc be the default for new users would improve accessibility

[exotic wasp]

you can't make replicated properties in bp for a uobject

why, idk

you can subclass an existing c++ class though

if you're doing something like an inventory system, I would make generic uobjects to hold state (like UFloatItemState) and subclass that to your item class (like UBatteryItemState) or something

[grand kestrel]

Because replication is tied to actor channels
When replicating UObject properties in C++ they get routed through an actor
I think Iris decouples this but IDK, haven't spent time looking at it yet

[exotic wasp]

right, but this is just a case of it not working in bp even though you can do it in c++ with no extra steps

feels like an oversight

[digital herald]

There are extra steps in C++

[grand kestrel]

Was doubting myself for a moment there, wondering if it was updated since I last had to do it lol...

[digital herald]

Yeah you still have to.

[thin stratus]

Stuff not available in Blueprints is usually not an oversight but on purpose.

[shrewd ginkgo]

https://blueprintue.com/blueprint/06aq24mf/
https://blueprintue.com/blueprint/ac3qzv7n/

steam overlay comes, I set up the server but when I search, the session result lenght is 0, it cannot find the server. someone pls help

[prisma knot]

Has anyone rolled out their own network for replication instead of using UE's default replication system?

[nova wasp]

it's possible but it depends on what you mean

you can roll your own entirely or you can use pieces of it

I have my own extensions of Iris network objects but I do not see the point of changing how it works internally

at least not for my purposes

[prisma knot]

I was just doing some research into UE's default Replication system and it's limitation with max players/performance etc.

[nova wasp]

I would check out using Iris

it should handle quite a alot more than the default

the big overhead of unreal replication stuff is mainly that it has to poll for some things and that a lot of work is per-connection

[prisma knot]

I see you said you've extended iris a bit, how is it using Iris?

[nova wasp]

also a lot of it is still 1 thread

[nova wasp]

It depends on what you mean

Iris for gameplay engineering is just a drop in replacement

[prisma knot]

Do you mean, enabling Iris is more of a "toggle" yes/no kind of thing?

[nova wasp]

Iris is the lower level networking backend and you just replicate properties and rpcs like normal

[nova wasp]

yes with some small config changes

you are better off reading the docs if this is news I guess

[prisma knot]

Ahh ok, I'll look into Iris and it's documentation then.

[nova wasp]

Iris replaces actor channels and is much more fancy as it is similar to raw arrays in C

the big downside though is that making custom serializers for it is VERY tedious

but that's not needed most of the time

[prisma knot]

I read about Iris a little while back, but It was listed as Experimental, so I didn't give it a good/fair reading into it.

[nova wasp]

it's quite close to being ready imo

but if it has issues you need to be experience with unreal engine source to mess around sometimes

Iris also kinda replaces the replicationg graph with prioritizers and filters

also no replays directly

it's ultimately just serializing bits

the part that I wonder about though is if you could make it do less work scaled by players

that is the big reason that unreal doesn't really just work for something like a 1000 player server

is because some work is per connection

that said it's not exactly an insurmountable problem... plenty of mmos ran on hardware from 20 years ago with more going on

but generally speaking unreal does not really fit being a huge MMO server in my mind

without changing something about the way it handles things per connection

or just heavily splitting up work on multiple servers (throwing money and complexity at the problem)

[prisma knot]

With that being said, can't that be mitigated by Sharding or Zoning based on server to alleviate the loads?

[nova wasp]

yep, ashes of creation kinda has the tech to do this

[prisma knot]

Yeah, I've played Ashes, it's not the most stable but it's making waves and getting there

Thank you for the insight though, I appreciate it. I'll start digging into Iris differences and documentation 😄

[nova wasp]

if you want to see an example of an mmo server made from scratch there's novus core https://github.com/novusengine

it scales to an absurd number of clients

I've never worked on anything with more than like ~60 player servers

so take what I say with a grain of salt I suppose

[grand kestrel]

I have worked on a couple of projects that did this
They always sucked to work on
And were never capable of client prediction

[prisma knot]

Interesting. Thanks for the feedback. Was it a mix of UE replication or full blown rewrites?

[tame sapphire]

Could I get some feedback on the following, I think ive made a match timer but would like feedback on improvements or changes

https://blueprintue.com/blueprint/485-_miz/

Func_Update Match Time

I think im more concerned around the multicast event in the blueprintUE upload

[nova wasp]

how often does this run

also why does it need to rpc if there's a replicated property already that can onrep changes?

[tame sapphire]

once a second, its for a match timer. I did want to do a second for a countdown that runs for 20s at a time when called and that would updated every 0.1s

it wasnt updating on client machines without the RPC so maybe i should be using the WBP for getting the time instead of updating via the RPC?

[grand kestrel]

I've seen both
I don't really have much input to add tho
Effectively they needed multiple servers simulating the same world

[tame sapphire]

on rep currently checks if the time is out of sync and if so updates accordingly

[dark edge]

There already is a synced network clock that probably works a lot better than this

It's not perfect but it's decent

[tame sapphire]

Im Guessing GetRealTimeSeconds (Did come across this in some other tutorials but there so many conflicting bits around I must admit I wasnt to sure)

[dark edge]

GetServerWorldTimeSeconds()

[queen cave]

are we allowed to let people test games or is it considered advertising

[nocturne quail]

250 A`Is in the level calling OnRep_EquippedWeapon(); which is used to replicate UPROPERTY(ReplicatedUsing=OnRep_EquippedWeapon) AWeapon EquippedWeapon;, which sets a value in the animation blueprint to get in Armed pose.
The issue is only 180 AI's pose updated and remaining have weapon attached in unarmed pose....

so the bool value didn't updated on all AI's, what can cause it?

[nova wasp]

are you sure it's actually not set?

if not it might be the server didn't think they were dirty if nothing sent it

if the animation bp is set during init

it might also show up after the onrep

in which case the thing that comes last should always read the value

[nocturne quail]

yes, I printed when onrep is called and it should be true because it prints success and still the AI anim pose not updated

[nocturne quail]

I did a work around by replicating a bool bisholdinggun instead of the weapon actor and everything works

seems like all bots calling a heavy object actor weapon using onrep is making server missed the events due to high latency or whatever not sure

[nova wasp]

I would recommend not replicating the weapon itself as a distinct actor object unless it is dropped if you can

it seems a bit easier to have it as like a subobject (or even just like, a weapon mesh + ammo in magazine or something)

but if it being an actor works that's fine

[nocturne quail]

New method works fine

old

void EquipWeapon()
{
    EquippedWeapon = SpawnAssignWeapon(bHoldingWeapon);
    OnRep_bquippedWeapon();
}

OnRep_bquippedWeapon()
{
    if(IsValid(EquippedWeapon))
    {
        currentaniminstance->UpdateArmedPose(bHoldingWeapon);
    }
}

New

void EquipWeapon()
{
    EquippedWeapon = SpawnAssignWeapon(bHoldingWeapon);
    OnRep_bHoldingWeapon();
}

OnRep_bHoldingWeapon()
{
    if(bHoldingWeapon)
    {
        currentaniminstance->UpdateArmedPose(bHoldingWeapon);
    }
}

[nova wasp]

instant crash when there's no valid anim instance

[nocturne quail]

it means there will be no breakpoint?

and the EquipWeapon is called when there is a valid animinstance

[nova wasp]

when you dereference a null pointer it will crash the program

unless this just omits a check

an anim instance could be missing during async loading or for other misc reasons

[nocturne quail]

server validate it before calling EquipWeapon()

and i logged on onrep if not valid, nothing printed

it means the animinstance is fine, if i decrese the amount of bots to 180 everything works fine

bot number 180+ didn't update anim pose

and all bots are childrens of one master bot class

[nova wasp]

this is an onrep

why would the server validate the client having a pointer

I am talking about what happens when this onrep is called on a client

you do not control that

you can kind of wait for init I guess but there's nothing stopping the client from losing the anim instance imo

[nova wasp]

are they budgeted? might be just capping out

[nocturne quail]

the question is why it is valid on other 180 bots?

[nocturne quail]

i don't know if they are budgeted, i just make many copies of them pressing ctrl+d in level

[nova wasp]

budgeted would mean using the skelelalmeshcomponentbudgeted and registered with the budgeter

about why the anim instances didn't update

[nocturne quail]

how to make then perma not budgeted?

[nova wasp]

it isn't unless it is

you can change the anim budgeted with some cvars, iirc there should be one to outright disable it

just not registering them with the budgeter should be it

and this is something you would have had to add yourself, this is not default iirc

they could also be just ticking based on visibility

but I dunno, you might have to just breakpoint the skeletal mesh component tick and track down who's doing it

[nocturne quail]

I will look into it, and one more test i did where i override the beginplay in each child bot class and called to reinit the anim class and it works fine on all bot classes

[dark parcel]

Do we need to do anything special for UObject that lives in a replicated actor to replicate?

Say I have a SomeUObject* someObj = NewObject
Will that replicate to clients?

[nova wasp]

https://dev.epicgames.com/community/learning/knowledge-base/7y39/unreal-engine-using-the-replicated-subobjects-list

ah woops I meant this one https://dev.epicgames.com/documentation/en-us/unreal-engine/replicating-uobjects-in-unreal-engine

[mystic estuary]

Hello, I have an actor changes its OnRep properties right before begin destroyed, literally the same frame. Does anyone know whether it's going to reliably replicate all of that to clients along the destroy event? Are there going to be instances when client will get it destroyed with old information?

[karmic briar]

MultiServerReplication going to change this.. someone tried using it and its pretty cool with ProxyServers and stuff.. MSR is similar to Ashes Of Creation MMO Server Meshing works..You can check out SabreDartStudios on youtube.. have some videos on MSR

[dark parcel]

does
#if WITH_SERVER_CODE

works in PIE?

[nova wasp]

I would not hold my breath

I would expect this to take years to be useful for the average dev

[karmic briar]

yup me too.. still experimental

[tardy fossil]

my guess is that preprocessor will always be true in PIE since the editor needs server code. UE_SERVER will always be false in PIE though, even if its running a dedicated server

[meager spade]

Fortnite has its own mesh servers

So epic have a working system already

Hopefully this is what they learnt from fortnite

@karmic briar

[karmic briar]

ah i didnt know that. i thought they just started since the MSR source code is quite barebone

hope that will be in engine in the future.. prob UE6 tbh

[karmic briar]

have some fortnite comment as well if i remember

right now MSR theoretically have have large players, just need to play with the proxy servers.. although each proxy server limits is still 100 players. what make this interesting is the illusion of large players. player from every server can see each others. so so like a single server can have lots of proxy servers and player can connect to this proxy server and can see each others

that why i said MSR is similair to what Ashes of Creation server meshing does.. at least in this aspect

since ProxyServer job is to relay data from server to client and vice versa

if youre interested, check out sabredartstudios on it.. really kewl tech

the 100 server limitations is still there per server

idk how fortnite does it but i think its similiar to this

[halcyon ore]

I'd 100% air on the side of client probably won't get it
IDK how exactly it works around destroying, and that such.
But, since UE plans for it to vanish from existence, so it may not even send the variable change.

[crisp shard]

if i want to have an actor give each player that's referenced in that actor the same widget, and have all the updates to that widget update to all of the references players what would be the best method in doing this?

currently, i just have the widget being created on each owning client, whenever it needs updated but each client is making their own widget instead of it coming directly from the actor. the actor just passes the variables, and doing it this way, i have to loop through all the clients to update each widget, and sometimes during bad connections some of those get dropped for the widget updates

it's like a "during game HUD" widget and it would update score, time, etc. so things that everyone would need to see but there's nothing personal directly related to any owning client

[weary jungle]

Just to clarify instead of having multiple widgets of the same type being created for each player you would like a single widget that each referenced player would access so you can change that widget and have it update to all players so one widget reference from your actor that is displayed on multiple players from your actor?

[crisp shard]

yea exactly

not even sure if it would be more efficient but seemed like a better approach, and might help avoid dropping some clients updates

[weary jungle]

Are you creating your widgets with bp or c++?

[crisp shard]

blueprint

[weary jungle]

I am looking into it and I don't think it will be possible to have one reference of the widget and display it on muliple players viewports will come up with slate errors for the shared ref with viewport registration. I think only way is you have to create the widget for each player based on the player controller which defeats the purpose. You could create some delagates to update the players widgets from that actor though.

I could be wrong

[crisp shard]

i was thinkning of doing it w delegates but realized i essentially be doing the same but with delegates

but yea, it does work and nothing that gets dropped, which is rare, is that imperative. if this was on the gamestate and was the entirety of the game, it would be slightly simpler but i am only doing it for the players that are involved in that particular game

[thin stratus]

if I only have access to a given PlayerController and nothing else, is there any way to check if it's the one of the Server? Roles seems to always be the same, as AutoProxy is more for Pawns. And I can't just compare it to the first PC in the UWorld Array.

Maybe I can check if the UPlayer is valid.

Yop, that worked.

[tardy fossil]

wouldn't HasAuthority also work? or would that return true on a client aswell

[meager spade]

Can always ask the player controller if it's a local controller

There's a function IsLocalController

Or listen server one then yeah that's a bit different

[exotic wasp]

IsLocalController and netmode check would work

unless you're trying to determine this as a remote

[weary jungle]

Ok so I have an idea it might be aweful but might work as sort of a hacky method you could try creating a single widget reference and then have a render target displayed of that widget for each player. They still have to create a widget but it would reflect changes from the one. Might not work I am not at home im at an airport atm just thinking about it.

This might only be good though for displaying, but perhaps not for interacting in the interaction case would prob need to be delagates

Also not sure how that would affect performance as well.

[mystic estuary]

But it needs to acknowledge the destruction

[rotund linden]

Hi, I'd like to explain my setup before the issue:
When my character overlaps with the box collision of a seat (which is a child actor attached to a boat), the seat sends a reference to the boat actor, the socket name, and the seat reference to the character. I then locally set these variables on the character, and replicate them to all using a Server RPC. After that, I also set a replicated variable InBoat? to true, which has a RepNotify.

Inside the RepNotify, if InBoat is true, I disable the character's collision and disable the Character Movement Component (CMC). I also set some boat-related variables and enable Ignore Client Movement Error Checks and Correction on the CMC. The reason I enable this is because otherwise the character sometimes teleports or jitters while inside the boat, due to movement correction, even though the boat movement itself is smooth. This makes it appear laggy.

Finally, I attach the character to the boat only on the server, using Snap to Target (with attach actor to actor but I also tried attach component to component). I think attaching is already replicated, that's why I do it on server (also tried to do it locally, and on all players). I also printed out socket name and some other values and they are correct.

The issue:
Occasionally, as seen in the video, the character is attached incorrectly or in a weird position. I don’t remember seeing this in standalone or listen server. How can I fix this?

[quasi tide]

I do an auth check + IsLocalPlayerController (for future reference)

[thin stratus]

Right. That owuld have been an option

Idk why I forgot about that

[quasi tide]

Looking at it - looks like you can even skip the initial auth check in some cases.

[gloomy mango]

Widgets should be controlled by a HUD which sits on each player controller. You should create delegates that listen to changes and fire off a function that will update the widget. If you’re looking for something that updated extremely often like time, I believe there are specific functions that are predefined that can help with that, you’d have to research that tho. But as far as most interactions you’d need delegates. (I work with C++ and I believe in blueprints they are called events.)
Also never use the tick function for any sort of listening for updates on widgets

I don’t see a point why there should be one widget shared between players unless that widget exists in the physical game world, each player should have their own widgets

What is your particular use case and intention for this widget?

I think you need to more ask the question of where you need to store the information that each players widget can use. For example, you can store round timer info in the game state as it is replicated down to all clients by default and you can easily access it from HUD to be displayed to each players widget

Basically you can’t get away with sharing a widget between players unless it’s actually in the game world as from what I understand widgets are not supposed to work that way

[nocturne quail]

AI is behaving weird, what can be the reason ?
on 1km map the pitch is fine, and on 32km map the pitch is dropping to 0 on client from what ever pitch is the current on server?

MAP 32km
LogBlueprintUserMessages: PITCH Server: -18.762428
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: -18.762428
LogBlueprintUserMessages: PITCH Client 0: -18.28125
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: -18.28125
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: -18.762428
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: -18.762428
LogBlueprintUserMessages: PITCH Client 0: -18.28125
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: -18.28125

MAP 1km
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0

void UArmaAnimationInstance::NativeUpdateAnimation(float DeltaSeconds)
{
    Super::NativeUpdateAnimation(DeltaSeconds);

    if (!IsValid(Character))
    {
        return;
    }

    Speed = Character->GetVelocity().Size();
    if (Speed != 0)
    {
        Direction = Character->GetCharacterDirection();
    }

    if (Weapon)
    {
        FRotator TargetRotation = Character->IsLocallyControlled() || Character->HasAuthority()
            ? Character->GetControlRotation() //keep server and local client synced
            : Character->GetBaseAimRotation(); //on other clients show replicated PITCH
        FRotator DeltaRotator = TargetRotation - Character->GetActorRotation();
        DeltaRotator.Normalize();
        Pitch = DeltaRotator.Pitch;
    }
}

[nova wasp]

Is the NPC getting net culled?

By the distance?

[nocturne quail]

we are closer, i can see him aiming on me but his pitch is going up/down continuously like some correction is happening

[nocturne quail]

directly doing this FRotator DeltaRotator = Character->GetBaseAimRotation() - Character->GetActorRotation(); without any local/server checks... gives the same result

[latent heart]

A rounding issue?

[nocturne quail]

not sure, because in level < 2km i don't have this issue

[latent heart]

Print the values in the log.

Every tick.

[nocturne quail]

already printed,

LogBlueprintUserMessages: PITCH Server: -18.762428
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: -18.762428
LogBlueprintUserMessages: PITCH Client 0: -18.28125
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: -18.28125
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: -18.762428
LogBlueprintUserMessages: PITCH Client 0: 0.0
LogBlueprintUserMessages: PITCH Server: -18.762428
LogBlueprintUserMessages: PITCH Client 0: -18.28125
LogBlueprintUserMessages: PITCH Server: 0.0
LogBlueprintUserMessages: PITCH Client 0: -18.28125

on server it is 0, and i can see it the AI is tring to render both 0 and the current pitch he is aiming at me

in result his pitch is jumping up/down

[dawn cargo]

So I am using Sublevels as a way to load subsequent sections/levels of a specific theme (each theme has 12 possible levels that 7 would be possible in any given "run").

trying to get the sublevel loading using a subsystem to control when and what sublevel to load and unload,

now I figured out that I have to RPC to client to get them to load the sublevel
however because the subsystem is replicated, where would be the best place/way to tell the client to load the sublevel.

I set the sublevels up in the editor so I am not making new Instances of the sublevels.

if this is a terrible way of doing this. what is a better way?

[latent heart]

Game state is a good place to send multicast rpcs.

[dawn cargo]

thank you

[nova wasp]

looks like something locally is fighting the value

track who changes it

[nocturne quail]

that value changes when bot starts aiming

since its an ai, and and pointing to target and i use his rotation.pitch to set animation rotation

also that issue is absolutely related to the level

because on other small maps this issue never exists

[weary jungle]

Is worldoriginrebasing true?

[nocturne quail]

this config is project related or level?

[weary jungle]

Level

@nocturne quail

[obtuse pivot]

How to develop an Axis in Runtime mode? There are few tutorials on this function.

[nocturne quail]

yes, p.EnableMultiplayerWorldOriginRebasing = true i have this in config file

[dark edge]

what do you mean by develop an axis?

[weary jungle]

Set it to false and test it on your larger map then tell me if anything changes.

[nocturne quail]

already did and no change

[weary jungle]

Ok interesting ill see if i can check it out a bit later for you when i can get to my computer

[twin condor]

Hello, I was watching a video guide on how to setup multiplayer for steam and was trying to invite myself on a different computer. When I accept the invite it kicks me out of my current session but fails to put me back into a session afterwards. I have a print string on my Onsessioninviteaccepted for the fail and success for joining the session. On Accept I get the failed print string, followed by a success print string, followed by another failed.

Here is a gif of whats happening

https://i.imgur.com/yUbABhy.png

[thin stratus]

Why is it printing 3 times? Is that 3 times the join node?

[twin condor]

so its printing the print string twice from the 2nd photo and one print from the success on the top photo

why its printing twice im unsure

i have that custom event being called only in the lvl bp for the empty level which is my LoadInLevel

https://i.imgur.com/5cLoASN.png

[twin condor]

When first launching the game the code from this photo runs a success but when i try to join my other account it is failing

[knotty bane]

isnt it always "listen?"

[knotty bane]

Dont really get the destroy session node there

[twin condor]

I'm unsure I just started messing with this type of stuff and was just following a youtube guide to set it up.

[knotty bane]

cause when you join you dont create a session, the hosts creates the session advanced

[twin condor]

yes but my on accept removes the player from the current session and should be joining the session that they were invited to. Or am I completely misunderstanding how that is supposed to be working.

or is there a different way for this to be done with me using 5.6

[knotty bane]

depends on how your game works normally you are in a menu and join a session, unless the player is already in a current session, but as long you are not the host there is no need to destroy session each time you accept and invite

are you using a server browser or only steam invites directly?

[knotty bane]

Its the same for most versions

[knotty bane]

there is no good or wrong it depends on what your are trying to make

[twin condor]

Only steam invites

If you have any guides that I could follow to possibly setup multiplayer I would greatly appreciate it. I wouldn't mind starting back from the beginning again if it works.

[sand dagger]

Has anyone had any success with Iris group filters in 5.6? I'm experimenting with team replication, only actors of the same team, represented as a group, should replicate to each other. The AddInclusionFilterGroup() / AddExclusionFilterGroup() and SetGroupFilterStatus() aren't working as I'd expect.

[royal pawn]

so would really appreciate a response on this.
i added a second capsule to the cmc
and i did it in the same way how the basae cmc dous it in prepmove
were the savedlocation = getcurrentlocation;
so that u dont have to have a seperate location variable u have to change each time u wont to move the capsule aswel
but im doing it with a childcapsule and im using relativeposition
and is getrelativeposition save to use in this context?
maybe the server and the client dont have the same result when using it etc?
or should i just make an extra variable for the location i keep track of

[tardy fossil]

i dont think CMC works well for anything other than a single capsule

[torpid crest]

any idea how to fix this?
[2025.06.29-13.52.45:322][364]LogNet: Warning: UNetDriver::ProcessRemoteFunction: No owning connection for actor BP_Character_Main_C_2147481933. Function DisableClientCorrections_Server will not be processed.
[2025.06.29-13.52.45:323][364]LogNet: Warning: UNetDriver::ProcessRemoteFunction: No owning connection for actor BP_Character_Main_C_2147481933. Function DisableClientCorrections_Server will not be processed.

[chrome bay]

Don't call the RPC unless you're the net owner of the object

It's just warning you that it will be dropped because you don't have permissions to call it.

[latent flicker]

What would be a good average byte payload size for a clients bullet shot? I just need something to go off. Right now I have about 30 bytes for a normal "shot" and within that its 23 bytes per "bullet" (shotguns or other weapons with lots of pellets cause me to have up to 12 bullets per shot.)

[latent flicker]

Are you familiar with what Hell Let Loose was sending roughly per client shot? I've seen you and others mention you worked on it previously.

I'm only sending damageable actors IDs the client thinks they hit, the direction, random seed and an some other things with each bullet. The server than of course validates this but I am using a trust and verify approach since playing anim instances on the server is costly and not in sync anyway so if I was to shoot an extended hand or leg it wouldn't align perfectly anyway.

[latent heart]

We send up to 21 bytes, but it can be less if it can encode is more efficiently.

[chrome bay]

ShotID, MuzzleOrigin, MuzzleDirection for a shot (reliable RPC)
Then for a hit, ShotID + FHitResult (also reliable RPC)

Client decided what limb etc. was hit. Not possible to validate that properly server side unless your playing animations on it AND perfectly syncronising all client anims.

[latent flicker]

Appreciate it, both of you 🙏

[chrome bay]

Biggest mistake people make with hit validation is trying to make it perfect. You're only trying to catch blatant cheating really.

Unless you've got an entire team to dedicate to it like CSGO or Valorant, I wouldn't bother. Also 100 players limits what you can do.

[quasi tide]

Yeah, I pretty much just validate on the server, "is this shot even potentially possible?" if I want validation. Also helps that I make co-op games only.

[steel anchor]

Hi, I'm using AddIpulse in multiplayer, but il only works on the server. The object has 'Replicates' and 'Replicate Movement' enabled, but the clients don't see the effect. Any idea why ?

[eternal canyon]

what are these RPC’s for

and why are you trusting the client with sending an entire UClass

as well as a transform

[steel anchor]

The reason I’m using a class is because I’m doing some parenting — it’s a parent object meant to spawn different child objects properly.

As for the RPCs: the first one is a Server RPC to handle the logic; the second one is an Owning Client RPC to get the client’s position, so I can know where to throw the object and get the proper vector. The last one is another Server RPC that actually applies the AddImpulse.

That last part is where I’m stuck — I thought doing AddImpulse on the server would work fine since the object has ‘Replicate Movement’ enabled, but for some reason, it’s still not replicating properly. ( the transform is just for where the object will spawn )

[gloomy mango]

I am pretty sure to handle steam invites you need to make your online code accomodate that, I am not exactly sure. From what I know it shouldn't just work like this without setting anything up

[gloomy mango]

This is most likely because of auto retry

[gloomy mango]

Actually now that I look, I am pretty sure you can't do this with steam ID for SPACEWAR

Do they do this in the turorial? I am skeptical that you can use SPACEWAR and features such as invites. I haven't tried implementing invites yet so I don't know yet. Right now I can just join session that is available on my game but haven't tried inviting

[twin condor]

Yes this use this is the demo

[steady onyx]

so, I'm using the Advanced Sessions plugin to implement online functionalities, but whenever I Join Session, the URL/Map the Client is trying Travel To, is not the right one (it loads back into the MainMenu)

[2025.06.30-19.15.01:242][873]LogBlueprintUserMessages: [BP_MainMenu_C_0] Joined!
[2025.06.30-19.15.01:242][874]LogGlobalStatus: UEngine::Browse Started Browse: "XXX.XXX.XX.X:XXXXX/Game/Maps/L_MainMenu"

I'm pretty sure it should be loading "L_GameplayLevel?listen", right? Does anyone have an idea as to where I went wrong with it?

[floral abyss]

Hello all. Had a question about releasing a multiplayer game on the epic platform made in Unity. Multiplayer is P2P with a matchmaking server used for initial connection. It's currently on steam, wondering if Epic has any policies for P2P communication for cross play?

[dark edge]

#epic-online-services

[timid nymph]

hey guys, very noob question.

My project is a c++ and blueprints project.

What is the best way to update the code of CharacterMovementComponent.cpp to always replicate the JumpMaxCount variable? It seems as though it only updates if you are the server.

Or is this all too much and I should just make a new replicated var through server RPCs to track this type of thing?

[fossil spoke]

That variable is actually marked to replicate in the Character class.

In ACharacter::GetLifetimeReplicatedProps

DISABLE_REPLICATED_PROPERTY(ACharacter, JumpMaxCount);

Its disabled though

If you override the Character with your own class you can probably re-enable its replicated state.

With RESET_REPLIFETIME_CONDITION macro

[timid nymph]

Could I remove this line?
DISABLE_REPLICATED_PROPERTY(ACharacter, JumpMaxCount);

or should i subclass the character class and add this line:

RESET_REPLIFETIME_CONDITION (ACharacter, JumpMaxCount);

[fossil spoke]

You are better off subclassing, you shouldnt be modifying Engine code unless you absolutely have to.

Be warned, replicating that property may have unintended side affects. Im not sure how the CMC would react to it being updated over the network.

If its been explicitly disabled, there is probably a good reason.

[timid nymph]

got it, so probably best to just create my own new variable to track it haha

[exotic wasp]

sorry if this wasnt what you meant, but property replication only goes from Server->Client

[timid nymph]

ah yea sorry i probably used the term wrong

I meant client calls -> server rpc -> multicast

[humble knot]

My boolean variable doesn't do an initial replication or OnRep for joining clients only if it was set to non-default in the editor details panel. (ofc the variable was changed on the server before the player joined). Is this some known bug or am i doing something wrong?

[chrome bay]

@humble knot It should be sent to clients if it differs from it's serialized value. E.g if the bool is false by default, and you changed it to true at runtime - it should rep.

If that's not the case that does imply a bug I think

[humble knot]

The problem happens if i manually set at editor time to true, then at runtime to false. The runtime change won't get replicated to joining clients and caues a desync state. If the client is already ingame everything works as normal. If i disable "Net Load on Client" it also works for joining clients, so i'll use that as a quick fix for now.

[chrome bay]

Yeah, definitely seems like a bug

[thin stratus]

That has been a bug for a long time.

If you have a property, e.g. a Boolean, that is set to X in the Blueprint, you then place an Instance into the Level and change the property to Y. If you change the property runtime to X, joining clients won't trigger the OnRep. The code that checks the value compares it to the Archetype for those initial calls and that will be equal then.

There is also not really a need to trigger the OnRep, tbf.

Cause you can just call whatever the OnRep calls on BeginPlay and it will update your Actor based on the current value of the property.

The assumption is that your actor, if the value hasn't changed, is already in the right state. So the actual bug would be that you aren't initializing the Actor based on the value properly.

Fwiw I also think it's an engine bug that it compares to the Archetype and not the instance. Pretty sure there is a UDN post that was opened years ago, where epic even explained it, but not sure what became of it.

@chrome bay ^

[chrome bay]

Yeah, I recall something similar on the forum as well but never got to the root of it

I so very rarely put stuff in the world directly that is replicated without defaults, maybe it's not just a widely tested case, IDK

[finite kernel]

ive ran into that 1 month ago when trying to replicate lightweight uint16 id for some objects
where the first free id was 0 which was also the default value 😄 thus the onrep never fired

now i just make sure to put them to some commonly garbage/invalid value

It was frustrating to find out

in case of boolean i would guess a workaround could be making it into enum with initial value

[snow island]

Hiya, glad to find this community! I'm pretty close to understanding replication in UE 5.6 with blueprints, but I'm having an issue now that I can't seem to resolve - I think it might be a misunderstanding in my head of how UE5 and Replication work.

I've been trying my best to solve this by myself for a few days now but I've been having no luck, would love if anyone could help please ❤️

My Setup is:

• Replicated Booleans for isSprinting, isAiming, etc
• I have an ENUM for movement mode that gets set inside the SetMovementMode function (Image 1)
• This function basically just contains other functions that will check the bool state and then update the movement modes accordingly, as well as setting a float called TargetMoveSpeed (Image 2).
• This function then takes that TargetMoveSpeed and calls a Reliable ServerRPC called SetMoveSpeed that then hooks up to a reliable Multicast for setting movement speed. (Image 3)
  This then calls a function called SetPlayerMovementSpeed that uses World Delta Seconds to lerp between the player's current MaxWalkSpeed and their TargetMoveSpeed (Image 4)

All of this basically serves to smoothly update the player's movement speed depending on their current movement mode. Idle is slower than walking, which is slower than sprinting, but when aimed, they should slow down a bit.

The behaviour works correctly on the server of course, but on the client, I'm seeing jittery behaviour, and the client's move speed being capped unusually.

Any advice please? Thank you! (Also sorry, posted then deleted/reposted cus i messed up image attachments)

[dark parcel]

@snow island there's no Blueprint solution for what you are doing.

https://www.youtube.com/watch?v=urkLwpnAjO0&ab_channel=delgoodie

^ for implementation and overview.

[snow island]

Thank you, somehow didn't find this one in my digging!

Unlisted, that's probably why haha

[dark parcel]

yea I don't know why. I used to be able to find it at the front page. Now I have to google it instead finding it from youtube.

[modest crater]

Yeah, doing what you're doing is going to have a bad time in BP land. It would be desync'd like crazy. You want to extend the movement components saved move and ensure the TValue is stored there and used for Phys move functions otherwise if you desync and it needs to reply moves it's going to use the current move speed for each move which is going to result in a different position.

To be honest a lerp'd speed is kind of really annoying to get right, set constants are better for this type of thing, it's why it's already set up how it is with just a walk and sprint speed and it doesn't lerp to it, it lets the cmcs acceleration handle approaching the max

[snow island]

Thank you both - feel like this has opened up a whole new world of stuff to learn UE wise now that I'm pretty comfy with blueprints

[jagged kernel]

Is there any way to replicate camera transform while using the smooth sync plugin on my character?

[dark edge]

why does anyone need to know where your camera is?

What are you actually trying to do?

[jagged kernel]

I am trying to replicate the line trace position things

The results are not correct when I use smooth sync. I think it is because of smooth sync just replicates actors. not components. But I need to know is there a way or not.

Or is there an easy way to use client authoritative network in Unreal Engine?

[grand kestrel]

Use GetViewLocation for location and GetControlRotation for rotation/direction
Edit: I don't think it gives you any kind of replicated view location that represents the camera's position

If you need control rotation on sim proxies then use GetBaseAimRotation instead

Typically if you need to do something from the view location away from the local client you're doing something that is misguided

[grand kestrel]

With Character/CMC yes
UCharacterMovementComponent::ServerShouldUseAuthoritativePosition()
bServerAcceptClientAuthoritativePosition
AGameNetworkManager::ClientAuthorativePosition

Last one is a global config so you can put in DefaultGame.ini

[/Script/Engine.GameNetworkManager]
ClientAuthorativePosition=1

[grand kestrel]

You can use my repo to either learn how to do this the CMC way, or just use the provided complete class -- which includes sprint, stamina, aim down sights, etc.
https://github.com/Vaei/PredictedMovement/

[jagged kernel]

Thank you a lot!

[bronze mauve]

Why do I get errors when calling SetIsReplicated from an actor component's constructor? Is this a no-no?

[quasi tide]

It is weird - just set the field directly

[thin stratus]

The error should tell you what to do

It wants a different call. Something with Default

[bronze mauve]

Doh! Thanks.

SetIsReplicatedByDefault(true);

for anyone who is interested

[quasi tide]

I've never used that method. I just set the field directly.

[grand kestrel]

Well, its not weird, it just doesn't like virtual functions in ctor

[dark edge]

don't just replicate the fact that you're casting the hook, replicate the hook direction and stuff

or maintain a replicated aim direction, Unreal already does part of that for you under the hood

[glad lotus]

Hello, could someone help me understand something:
I have a UAnimNotifyState that i'd like to use to locally simulate some effects that are not game breaking if simulated incorrectly, so there is no point of doing any sort of replication for those.
But these notifies work for the autonomous proxies and do not seem to be triggered at all for simulated proxies, is this something that's just not supported? Or am i missing some checkmark somewhere?

[simple remnant]

Could someone tell me why when I try to load up a listen server I can't control the player but when I have it set on one player I am able to control the player?

[nova wasp]

Not without knowing anything about how you bind input

[weary jungle]

Might be due to posession of the player controller

[simple remnant]

I fixed it

it was just a setting but now I have a replication error as when i sprint the server shows me moving in place

[weary jungle]

Which setting was it if you don't mind me asking?

[simple remnant]

i had auto possess player on player 0 so i set it to disabled

[weary jungle]

So was a possession issue. Also for the second issue is it only when sprinting?

[simple remnant]

everything in general

[weary jungle]

Is the character moving at all with or without animations?

[simple remnant]

with animation

it just doesnt go in the direction im moving to

[weary jungle]

Is this only on the server or both the sever and the client?

@simple remnant

[simple remnant]

the client i cannot move

but server i can

it shows no idle animation of client from server

[weary jungle]

Do you have bReplicatesMovement set to true?

[simple remnant]

is that in the third person character or the cpp

[weary jungle]

You can do it in either the characters blueprint or a cpp character class

[simple remnant]

yeah its enabled

does the other options matter below?

[weary jungle]

One moment checking something out

Ignore what I just said because you had it on 0 before

[simple remnant]

👍

[weary jungle]

Do you have a specific gamemode set?

@simple remnant

[simple remnant]

wait

its just bp_gamemode

[weary jungle]

^^

Is it working now

[simple remnant]

sadly no

is it an issue with the cpp?

[weary jungle]

In the gamemode do you have the bp class or the cpp class set?

[simple remnant]

sorry will check later busy atm

[gloomy mango]

Question to you guys. @ me or reply to me so I see this tomorrow cuz for now I’ll just go to bed lol. Is there a thing like steam soft timeout when you call findsessions too much in a very short succession? I think I might’ve accidentally did that (it hasn’t been an hour yet and idk if I am timed out how long it is) but GPT suggested that I did in fact got timeout because I did kinda spam my find sessions button and it finds like a crapload (1000+) (yes I know it was dumb, I’m lowering it down to like 25). But my point is I haven’t changed any join/host code and suddenly sessions are failing to be searched where previously I was able to find my test session on another machine in a packaged game

[simple remnant]

It's a blueprint class @weary jungle

But I don't think thats the issue. I have a custom cmc set up so I'm not very sure where the problem is coming from.

[thin stratus]

One thing could be that you are mixing GameStateBase/GameModeBase and GameState/GameMode.

That has nasty side effects. But log would be throwing a warning at you if that's the case.

Other than that, how are you spawning and possessing those characters?

Feels like you might be possessing it locally.

[simple remnant]

Server 0 can move but client 1 cannot. Client sees server 0 moving but only in place and Server 0 sees client 1 completely still. And my animations doesn't follow which direction I move to.

[thin stratus]

Why do you select an AI controller as PlayerController

That's not correct

Also for multiplayer you might want to switch over to using the non Base version of both GameState and GameMode. Won't be the reason for your problem but the Base versions are more for simpler non multiplayer games

[thin stratus]

Moving but still meaning the location changes but the animation doesn't?

Can't view the video, sitting on a bike in the gym atm

[simple remnant]

When I switch over from the non Base version I cannot control my third person character

[verbal ice]

Is there a way to open a special channel to transmit files to clients

[nova wasp]

there are sockets and you can send stuff over http from what I understand

if it's not huge you can just yolo an rpc across too

[verbal ice]

Yeah RPCs would work but not ideal

Was thinking like the Source engine's ServerDL stuff

Where players can download server content before fully connecting

Although maybe I can have the dedicated server host a web server itself for transfers

[nova wasp]

not sure how I would do it, it would depend on how huge they could get

[verbal ice]

but that's not really ideal

[nova wasp]

it seems kinda sad to force the host to upload a billion files but hosting it yourself is not so nice either

[verbal ice]

wdym

[nova wasp]

bandwidth upload speeds would can be quite restrictive

that said I suppose it could be capped low

[verbal ice]

Source caps it low yeah

That's why hosts have the option to serve content through a web server

Games like Garry's Mod still serve scripts through the game server though

I wonder if I could like

add custom net control messages

Might have to do that if I want to do stuff like sending scripts during the connection process and not when the player's connected

[sullen sail]

Hi, I’m working on a basketball game and using a Layered Blend Per Bone node in the AnimGraph.

I have 3 animation layers:

Main movement state

Upper body defense pose (Layer 0, applied at spine_01)

Head rotation (Layer 1, applied at neck_01)

The problem: when my character enters defense mode (activating the upper body blend at spine_01), the head rotation (on neck_01) stops working.

Does Layer 0 override Layer 1 in this setup?
If so, how do I make sure the head rotation still blends properly while the defense pose is active?

[thin stratus]

You have to switch both

GameMode and GameState

If you have BPs of either already you simply reparent them in the class defaults of each BP

Your screenshot had only a custom GameMode, so reparent that to GameMode instead of GameModeBase and select the matching native GameState class for the... GameState class

But that's not solving your initial problem

[simple remnant]

could it be a cpp issue?

[thin stratus]

What exaxctly?

If you have code in C++, then yeah of course. This is a user bug after all.

[thin stratus]

Shouldn't there be a multiplier for the percentage of blend, aka the weight?

Also this is an #animation question. Why is this in Multiplayer?

[thin stratus]

--

Are there any known ways to display debug information (specifically debug rendering) on Clients if the data exists only on the Server?

[nova wasp]

yeah, unreal does this already for some things

[thin stratus]

Yeah I also remember having seen this, but it's hard to search for.

[nova wasp]

AGameplayDebuggerCategoryReplicator?

vlog is also nice for this as it can split per world in the editor

[thin stratus]

It needs to run in builds, so can't have it use Editor only stuff.

And what I remember is apparently this:

void DrawDebugLine(const UWorld* InWorld, FVector const& LineStart, FVector const& LineEnd, FColor const& Color, bool bPersistentLines, float LifeTime, uint8 DepthPriority, float Thickness)
{
    if (GEngine->GetNetMode(InWorld) != NM_DedicatedServer)
    {
        // this means foreground lines can't be persistent 
        if (ULineBatchComponent* const LineBatcher = GetDebugLineBatcher(InWorld, bPersistentLines, LifeTime, (DepthPriority == SDPG_Foreground)))
        {
            float const LineLifeTime = GetDebugLineLifeTime(LineBatcher, LifeTime, bPersistentLines);
            LineBatcher->DrawLine(LineStart, LineEnd, Color, DepthPriority, Thickness, LineLifeTime);
        }
    }
    else
    {
        UE_DRAW_SERVER_DEBUG_ON_EACH_CLIENT(DrawDebugLine, LineStart, LineEnd, AdjustColorForServer(Color), bPersistentLines, LifeTime, DepthPriority, Thickness);
    }
}

But that stuff is Editor only too, which makes sense ,cause the Server has access to the Client Worlds. PIE is the easy thing thing here.

Will set up my own thing then I guess and will check the AGameplayDebuggerCategoryReplicator. Thanks!

That actor looks interesting. Thanks for sharing!

[lament cloak]

Hey all!

So I'm building a custom vehicle using Chaos Modular Vehicles as a base. I've also built a custom simulation component for levitation (think Star Wars Pod Racer).

It feels really good for the controlling player, and impacts and shit seem to be all in sync. However, simulated proxies look pretty bad. It might be because of the speeds involved?

I see a lot of skipping and teleporting. Ping is ~60ms and stable, 1% packet loss.

Things I've tried:

• Increased min net update rate to 60hz on the vehicle (pawn)
• Set prediction mode to Full Re-simulation
• Set Simulated Proxy Network LOD to Forward Predict
• Increased fixed tick to 128hz (and then set it back to 60hz)

Questions:

• What is Network LOD? Is that Level of Detail?
• Does my custom simulation component (inherited from UVehicleSimBaseComponent) module need Chaos::FModuleNetData and utilize custom net serialization? All I'm really doing is taking in the user's throttle and steering and applying forces... I also have no idea who precisely this data gets replicated to, is it for simulated proxies? Because for my own vehicle, it feels fucking amazing. So I'm not sure if I really need it or not.
• Does anyone know what exactly Chaos::FSimOutputData is used for? I can't seem to understand this area of the code and why it exists. Maybe for debugging purposes only? I cant seem to find it being used anywhere, I'm now thinking there may be some reflection involved and therefore hard to grep.
• It is unclear to me how simulated proxies are receiving data from the server. Like, are they receiving the controlling player's inputs too, or is it just receiving velocity, position, etc?

I understand all this shit is experimental and/or in beta, and quite advanced but literally any pointers would be helpful!

Also bonus question, does anyone know why the fuck there's so much networking config under the Physics section? And not under the Networked Prediction section? gah

You can see the teleporting on the right viewport especially near the end. Both of these are clients.

Bonus question 2: What are those visual artifacts on the ground? annoying

[fallen fossil]

Net Load on Client,
is this relevant for joining players?

[chrome bay]

No. It means clients won't load the actor locally, and instead will spawn it from replication.

(if replicated)

[lament cloak]

Fuck yes I found the solution to my problem above for networked physics simulated proxies. You add a NetworkPredictionSettings component and set "Focal Particle Physics Replication LOD" to true

[snow island]

Just wanna say thank you again for this and wanted to 1000000% recommend this series to everyone else who has even a crumb of interest in dipping into C++, I watched the full series last night and it's a lot to take in at first (and I will likely need to go back) but really, this youtuber is great

https://www.youtube.com/watch?v=urkLwpnAjO0&list=PLXJlkahwiwPmeABEhjwIALvxRSZkzoQpk

[fathom mural]

Hey Guys! I'm very new to replication systems
Can anyone give me a hand on a Networking issue I'm having? I'm trying to use replication on a weapon inventory system, but for some reason, one of the variables is not being read, only by one client, and the variable IS being read and used on the character's anim blueprint without a hitch

[nova wasp]

how is it replicated?

[fathom mural]

Flagged as RepNotify and set in a Run On server event

To clarify
I have "CurrentWeapon" marked RepNotify in my InventoryComponent. When a player equips a weapon, I call a Run-on-Server event to assign it on the server. Then Client PCs receive it via OnRep_CurrentWeapon, where I use a direct reference to the player (also replicated IN the component) and set a replicated EquippedWeapon variable in the Character Blueprint

[nova wasp]

what is the InventoryComponent owned by?

[fathom mural]

By the Character BP

[nova wasp]

you are going to need to be more specific about what client is missing the replicated property

are simulated proxies not receiving it? just autonomous?

also " direct reference to the player" could mean many different things

also there is probably not much need to replicate the character inside of a variable in the inventory componenent but I am having trouble following

the character is already replicated, the component already has them as the owning actor...

also no clue what type "replicated EquippedWeapon variable" is

is it an object? an actor? an integer?

[fathom mural]

Believe me I have no Idea of what I'm doing, it's been three days since I started to use any replication at all.
When I start the Gameplay I set it to two players, and set it as a listenable server***, the second player is the one not getting the value that I need. The reference to the player is set on the component's begin play by getting the owner and casting to character's blueprint and storing the value. The way the inventory is set up (very likely not the best) is each inventory component has an array of weapons (object reference to weapon BP). When a weapon is swapped, I send an event to the inventory with an index to set current weapon. Let me see if I can take some screenshots (im working on Bp only for now)

[nova wasp]

the second player is a client

you are better of reading the pins

[fathom mural]

This is the swap-to event. The run on server event i only use if the player is not locally controlled and doesn't have authority

On the OnRep for tha value I have it set up like this

And for some reason, this is the function where it reads None, but the blue print bind does read it normally.

[nova wasp]

just because a property is replicated doesn't mean it represents something that can replicate

I have to go but I think you might be trying to replicate objects that aren't replicated

[fathom mural]

The weapon actors are set to replicate and so is the inventory component

[nova wasp]

replicating raw data like numbers and enums will work as long as the thing they are in replicates, but a reference variable to an object assumes one of two things

• it is network stable like an asset (for example, a static mesh asset)
• it refers to a replicated object

also there are two different current weapon variables here and I am confused

[fathom mural]

What I don't understand is how the client can read the "Current weapon type" value i'm setting from the current weapon properties but not the current weapon itself

[nova wasp]

read my message again

an object is not simple to send

you are not just sending a number here

https://dev.epicgames.com/documentation/en-us/unreal-engine/replicating-object-references?application_version=4.27

which is of course not available for 5's version of the docs... I would say it's safe to assume this is largely identical in 5

[fathom mural]

I think I understand, but I still don't know how i'm going to fix it Lol

Thank you

[nova wasp]

I can't say much just seeing random nodes without seeing how they fit together

or what the weapons actually are

I have to go, hopefully someone else is around

[fathom mural]

No worries, thanks either way

[nova wasp]

I appreciate the effort in sharing the info I guess

an onrep is a good idea here because it's persistent and gets called when the value changes

might be useful to show how the gun actors are created

[fathom mural]

Got any tips on sharing blueprints without having to take awkard screenshots? hahaha

[blazing spruce]

https://blueprintue.com/

[crisp shard]

When switching pawns during runtime, are there any approaches to making the camera switching feel a bit better? I find that the snapping to new pawn is just a bit jarring but maybe I could just add a simple widget to overlay while it’s happening but it’s so fast that I’d think like a screen fx or something would probably be more of a useful solution. But yea not sure what options there are to smooth out that operation

[teal heath]

How can i make an inventory (array of three slots) thats different for each character and to work in multiplayer? (like when each person picks up a new weapon it gets added to their inventory seperately)

[modest crater]

You mark the array as replicated and ensure the actor/component who owns it has replication enabled

[teal heath]

so if i make this replicated will it work?

@modest crater

[modest crater]

I think you should really take the time to read some of the pins in this channel like the Network Compendium https://cedric-neukirchen.net/docs/category/multiplayer-network-compendium/, trying to implement networking with no real idea sounds like a painful time.

[teal heath]

i have an idea i just don't know how to make it work

[sick estuary]

Hello everyone, has anyone experienced the following issue? I've been encountering this recently in my multiplayer UE5 game.

When a player dies, they are possessed by a SpectatorPawn. Upon respawning, they are re-possessed into their Character.

The issue is that sometimes the server (host) cannot see the respawned client's mesh, although all other clients can see each other's meshes without any problem. Collision still works, and the Character actor exists—only the mesh is invisible from the host's perspective.

Has anyone come across this or have any ideas on how to fix it? Thanks!

[ionic sapphire]

Hi, does anyone know how to go about getting a client to load in a map from a pak file? My sessions are currently being setup very closely to how they are in the example in the Network Compendium, but when the client tries to connect to the host's IP address in the TryTravelToSession call, it loads into our main menu instead of the map the server is actually on. This wasn't an issue before we started using pak files, but I'm unsure what I need to adjust to make it work. Appreciate any help you can provide on this 😅

[proper pagoda]

Usually if it opens the main menu (or whatever is your default map), this usually means that a step in the loading process (or even post loading) has failed and/or that the connection has been kicked out. You should look at the logs to find out if it's that the load has failed entirely or if the connection is being closed by the server (usually with a CloseReason alongside it).

[worldly needle]

How can I troubleshoot in multiplayer game when Character Movement Component fixes location of my character in code side ? Any documents anything you know ? I would be really appreciated.

[ionic sapphire]

Ok so I'm getting a ConnectionTimeout as the CloseReason in the client's logs. Are there any common reasons it could be timing out? Anything pertaining to loading from pak files that I could have missed?

[ionic sapphire]

Sorry, my message was meant to be a reply to this

[proper pagoda]

Honestly I'm not sure. We use .pak files in our project and I've never had connection timeouts due to them. If you're certain it's related to this (meaning you've done A/B testing with solely .pak files on or off) it would be beyond the scope of my knowledge.
In the logs, before the timeout you should check what the client and or server is attempting to do that repeatedly fails or just never happens

[ionic sapphire]

As far as I can tell, it finds the server's IP address just fine, but it tries to browse to the Main Menu Map as if it thinks the server is on the main menu map, despite it being on a different map entirely. I'm providing my client log here. It's connecting to the server around line 904, and it times out by line 993.
I have confirmed that this is solely related to us switching to pak files, but that's about all I've confirmed. I appreciate your help thus far nonetheless. Even just finding the ConnectionTimeout is further progress than I've been able to make the last few days.

[proper pagoda]

Wow that is strange indeed

[ionic sapphire]

Well at least I'm not crazy lmao

[proper pagoda]

Usually what happens during a connection is that the client sends a connect request, and the server responds with the Map (if it accepts the request)

[ionic sapphire]

Is there any common logs I can search on the server logs to see if the client connects to it? I can't immediately find anything on the server side.

[proper pagoda]

You could try and log what SETTING_MAPNAME is used for the session. And also what map is the server opening with the ?Listen argument

[knotty bane]

hey guys been trying to make a similar system like chained together but i keep glitching out these characters, can you see where im going wrong since my phyics constraint is not working at all

[dark edge]

Why would the CMC care about the chain at all?

[knotty bane]

wym?

[halcyon ore]

There saying.
Why would the CMC know of what a "chain" is/ thus care about it.
CMC isn't physics based, so.

[knotty bane]

but where am i going wrong since players are not constraint and bugging out?

[swift violet]

I'm having the same problem too and it has nothing to do with the CMC. anyone know if there is an alternative to what he does?

[formal solar]

Set view target with blend. But idk why you posted in multiplayer section this is irrelevant to MP

[proper pagoda]

If your pawn uses the the CMC then its movement isn't operating with the same system as the physics system. The physics constrain component is explicitly made for the physics system. It makes sense in that context that the players aren't affected by the constrain. However it looks like the constrain has some other issues in the video.

[knotty bane]

ah, but how could i constrain 2 players together cause im lost at this point

[proper pagoda]

For multiplayer, without going through C++ I wouldn't know exactly.
Although, constraining 2 players is somewhat simple if the only rule is that they must remain within a certain distance of each other, as long as the max distance is replicated you can just enforce it in the movement function.

However if you want something closer to chained together it might be worth to look into replicated physics and physics based movement

[knotty bane]

thanks for the answer, but if you see the bp's am i doing something wrong since the constrains are not doing it? the chains are not really a problem

[proper pagoda]

Your constrain is trying to constrain the two "Mesh" components together, however since those components aren't simulating physics, it cannot actually enfore anything

[knotty bane]

ah

Ill look into that thanks!

[gloomy mango]

Hey people, is there some good way to make sure when you testing session search functionality on APP ID 480 you get back consistent results all the time? Lets say if I change my session name in session settings to something very specific, will I then be able to SearchSessions(5) for only 5 results and consistently get back the result set?

Reason I ask is because yesterday I did an oopsie and might've ended up getting timed out by steam cuz I sent too many session search requests cuz I was under the impression since I am testing on ID 480 I'd need to search far and wide to get my actual session. And I want to find some consistent result that doesn't involve sending crazy session searches since I am testing some multiplayer UI functionality rn (like looking for available lobbies widgets)

[simple remnant]

I had written my cmc so that could be the issue

[thin stratus]

Probably not. Too many others are using this AppID. Your best bet is to simply get your own AppID.

You could try to set a special ServerSetting on it and then use as special QuerySetting that checks for that.

That might return it on Steam's end

But if that fails then you just gotta live with getting your own AppID

[gloomy mango]

I literally don't mind paying $100 if it wll allow me to just use that custom ID for development without having to make up a store page or anything like that

[thin stratus]

Store page is only needed when you releases

[gloomy mango]

Alright, so I can basically submit no code to Steamworks but process a custom ID and use that?

[thin stratus]

You can use the custom ID and you can submit builds to Steam, all without having to release or do the page yet.

Most studios have automation running that pushes builds to different Steam branches that are available to everyone added to the Developer group in Steamworks backend

So they can test the builds

But you can also ignore the builds stuff for now fwiw

[gloomy mango]

Alright cool, I wasn't aware of this workflow, I will look into this, thanks a bunch

[thin stratus]

You do need to add a second account to the devs group though, cause you gotta "own" the game to play it

Otherwise you won't be able to host and join :D

(hard to do that with just 1 player after all)

[gloomy mango]

So just add my secondary steam acc I use for testing to the devs along with mine

[thin stratus]

Basically, yeah

The Steam backend page is the most confusing shit there is

But I'm sure you'll manage

Lots of others did :D

[gloomy mango]

Thanks for the info, I just needed to know where to go cuz random spotty connections was starting to get me very annoyed

I tried using SearchConnections(10k) to circumvent that but apparently Steam didn't like that xd

It worked... FOR A BIT

I think I got timed out yesterday, I will double check today should be good but I am really testing the online code so my own app ID with like SearchSessions(5) temporarily will help a ton in avoiding steams spam logic

[quasi tide]

It is so damn confusing. Wish they would actually address it. But they never will.

[thin stratus]

You can in theory do almost everything you want just locally with subsystemNull. Only if you really want to test steam you need to use steam

[quasi tide]

I have like 5 projects on Steam. I still get turned around and lost.

[gloomy mango]

Yeah I don't want to have local work only for nothing to work on steam side xd

[gloomy mango]

Document EVERYTHING you do. Its what I do lol

I have a literal physical notebook

It kinda helps

[gloomy mango]

Btw congrats on 2 releases! Thats wild!

I checked them out look cool!

[tame sapphire]

Would any one mind expanding a little on how to use GetServerWorldTimeSeconds() for a server/map network time?

As in it seems like I would need to call a game state event/function on tick to update (Just wondering if theres a better way)

Widget :

Game state function :

I imagine i could start a timer in the widget and loop it every 0.5 - 1 second which is fine for seconds (Or so i imagine) but I want to do a turnbased decision system and you get 20s for that timer and i would ideally like to show milliseconds so wondered if tick is ok here as its only 20s every 1.5 - 2 mins or should i go the timer route and just loop it every 0.01s?

[proper pagoda]

Calling a getter on tick is trivial in terms of performance cost, so it should be fine

[proper pagoda]

Keep in mind that GetServerWorldTimeSeconds returns approximatively the time the server is at when they sent the update. Technically by the time it reaches you, you're about half ping ahead

[tame sapphire]

Thank you, however just to be clear would a timer be better instead of tick so its only called when used for the ready time? the match time would run constantly so could see it being on tick but i don't think ready time should be but im guessing in the dark?

[proper pagoda]

If you make sure the timer runs at most once per frame then it would be better, but kind of in the same way that standing on a sheet of paper makes you taller.

[tame sapphire]

@proper pagoda understood and ok great thank you

[glass plaza]

Is there any way to start a Session without having one of the clients the Server? I would probably have to start a dedicated server?

[formal solar]

Dedicated servers are either very complex and time consuming or very expensive

[fossil spoke]

I disagree. They CAN be expensive, but that all depends on your game and where you choose to host.

Not sure how Dedicated Servers are more complex than Listen Servers though.

[fossil spoke]

Sessions dont really care what type of Server you are running.

A Listen Server has a Host.

A Dedicated Server does not.

Therefore if you want a Server without a Host, you need a Dedicated Server.

[formal solar]

Expensive for an average joe

[fossil spoke]

I still disagree, there is nothing stopping you from using your own Hardware to host a Dedicated Server.

[formal solar]

That's pretty CPU intensive right?

[fossil spoke]

Which is probably more expensive than just renting something like an EC2 instance or even going more bare metal with a less elaborate hosting provider.

The point is, there are cheap ways to host Dedicated Servers.

It all depends on your game and your needs.

Scaling Dedicated Servers can get expensive, but again, thats relative to how expensive it is for you to just host 1.

[grizzled stirrup]

Do you happen to know of any resources that show how to automate and upload to steam in one step? I have packaging automated and the steam uploading automated but it's still a manual copy paste job to put the packaged files in the correct places to upload

[grizzled stirrup]

Requires a machine running at all times for starters. Doesn't scale without you spinning up more machines. Listen scales infinitely depending on the userbase

[glass plaza]

Thanks! Architecture change time, I guess

[fossil spoke]

None of that was part of the original question. He didnt ask which option was more or less complex to setup, nor did he ask which was more or less expensive.

Eitherway, I covered scaling in my response.

[grizzled stirrup]

Fair I just read your sentence to be fair 😄 I'd say they are also more complex just because you also need a source build and need a separate client and server exe

[fossil spoke]

Sure, but from the point of view of the original question, complexity of setup was not really part of it. Sessions for both Dedicated and Listen Servers are basically identical.

You are correct though

There is much more that goes into a Dedicated Server than Listen Servers.

[halcyon ore]

Any tips, or videos on how to optimize network size of RPC events?

[fossil spoke]

Depends on what the RPC is for and what Information you need to send through.

Generally speaking you want to ensure you are sending the minimal amount of information you need.

This is very specific to the RPC as well.

Optimizing anything can be tedious.

[halcyon ore]

Right now its not anything crazy, I have kept it pretty basic for my needs.
But, even now it feels a bit high, for the "basic" data i'm sending back and forth.

Class
int64
2 FString's
Array of Fstring
Array of Int32

[fossil spoke]

As I said above, optimization can be really specific. I have no idea what any of that means relative to your use case.

For all I know you might need all of that information and there is no way to optimize it.

[dark edge]

What's the actual thing you're syncing?

[fossil spoke]

Yeah, if you could explain what it is you are trying to do and why you think you need all of that, we might be able to help you consider some different approaches for cutting down on bandwidth

[halcyon ore]

Its for replicating data outside of network range.
So, there could be 1 entry of this, right next to the player, or 400 spread out everywhere.
So, I find it more "logical" to just RPC all 400 entries no matter where they are, so its all 1 shared handling.

The data is:
The class of the actor
Its custom ID, defined by me.
Its custom name
Its custom UI based description of it.
Then, the 2 arrays which are my method to replicate a TMap
Of various "generic" data. (not needing to be an explicitly handled variable, so I can just run a "find" on the TMap

The only optimization that I could think of, is that not all the TMap data needs to be int32, so maybe I could make like 4 different versions, or each int type?

For instance, I store 1/ 0's in the Tmap, just because its easier to run a find, rather then handle the shit for adding a new variable, personally.

[dark edge]

how often?

[halcyon ore]

Only main usage is on UI open.
I have RPC's setup, that handle just the data that changed, if data does change while the widget is open, so there far smaller, since its at most 1 int32 and FString

[dark edge]

Why not just request on UI open

or if it's not changing that fast then maintain all the data with normal replication

fastarray and such can just send deltas

[halcyon ore]

1. Thats what i'm doing? (requesting on UI open, but its 400 entries, let alone if I add more data, since i'm still working on it)
2. Wouldn't normal replication still have issues, if all 400 at once is too much?
3. I assume fast array is related to #2, because idk. 😛

[dark edge]

When do you need to view 400 at a time?

I think your best options are to either maintain synced data or just request what you need when you need it

[gloomy mango]

By the way, so turns out I can just add a custom key like this:

on CreateSession logic:

LastSessionSettings->Set(FName("DevSessionTag"), FString("YourUltraCustomComplexStringHere"), EOnlineDataAdvertisementType::ViaOnlineServiceAndPing);

where LastSessionSettings is TSharedPtr<FOnlineSessionSettings>

and in FindSessions logic.

LastSessionSearch->QuerySettings.Set(FName("DevSessionTag"), FString("YourUltraCustomComplexStringHere"), EOnlineComparisonOp::Equals);

where LastSessionSearch is TSharedPtr<FOnlineSessionSearch>.

Then you can straight up specify your find session logic like this when looking for lobbies MultiplayerSessionsSubsystem->FindSessions(10); and it will work as just 10 results even for AppID 480 since now it will have an extremely specific filter in both hosted sessions and search queries.

This will avoid Steam shadow time outing you in case you are looking for a bunch of sessions on FindSession cuz of AppID 480

[dark parcel]

Any specific reason to use appID480?

[gloomy mango]

Its the free one everyone can use and not pay $100 to register your own

But as a result a lot of devs use it to test

[dark parcel]

Yeah, i mean when you want to release your game you will have to cough that 100 anyway

[gloomy mango]

So you might not find your session without specific key filtering on doing something ridiculous like FindSessions(10,000)

[dark parcel]

Yeah you will end up seeing other game project session.

[gloomy mango]

I was not aware of that and got shadow time'd out yesterday, fix was simpler than I thought

[dark parcel]

I dont bother with appid 480 since i want to have ny own environment

Not to mention its inevitable to pay if you want to release

Also it sucks having steam pop out space war instead our own game name.

[gloomy mango]

Well if you want to just test multiplayer behavior nothing really changes if you use your own or 480 with custom filtering

[dark parcel]

Certainly thats what the 480 is for.

[gloomy mango]

Well yeah, but this way I can pay this in 2 years, or 10. . . or 20. Or ever???

Not that I don't have the money, I don't mind grinding spacewar anyways

lol

I am in very early stages of my project I just decided to do multiplayer stuff first

[dark parcel]

Yeah back then when i did some testing, i just get all the session

But only push to U.I the session with special key

Just a setting / mode that my specific project specify to mkae it unique

[halcyon ore]

I wonder if "Advanced Sessions" is broken, because I do this same thing, and it doesn't work. 😛

[gloomy mango]

I don’t use any plugins outside of SteamSockets and SteamOnlineSubsystem I have my own session logic

[thin stratus]

The Steam upload scripts just have to point where you packaged the game to.

[hollow swallow]

Do overlap events only happen on the server? my client is never getting them

[nova wasp]

overlap events are from the a local physics query declaring an overlap

unreal runs on both computers in a similar fashion, physics/collision does not only exist on the server

[hollow swallow]

okay thanks, for some reason only the host is adding enemies on overlap with sphere and client isnt :/

[nova wasp]

you might want to consider if the sphere is the same on the client at all

or even present

(or the things you expect to overlap with)

[hollow swallow]

Yeah its the same on both, nothing changes it, the server spawns the "spell" which has the sphere on it to get the nearest enemy

[nova wasp]

the server spawns the spell is not the same though

that is replicating it

is it being spawned on the client or not?

remember: the client is a distinct world

[hollow swallow]

Nah only the server. I thought only the server should spawn actors

[nova wasp]

by this I mean does it get received by replication on the client?

[hollow swallow]

the actor i spawn is set to replicated, so i thought so?

i can see the spell on the client

[nova wasp]

when it appears on the client, is it in right position to overlap the other body? do the bodies both have the overlap settings needed to trigger an overlap?

with this stuff timing can get weird

so it might be just moving too fast etc past something

[hollow swallow]

yeah spawns in the correct spot and everything, it doesnt actually move yet, just spawns on player location, and should get enemies within range of the sphere scaled by 20

[nova wasp]

if it doesn't move when would it update overlaps?

[hollow swallow]

ill try a reboot, have had it open a long time maybe its just bugging out

the enemies move towards the player at the start

they walk into the range

[nova wasp]

https://dev.epicgames.com/documentation/en-us/unreal-engine/collision-in-unreal-engine---overview

[grizzled stirrup]

I guess I need to figure out how to upload automatically when UE is finished packaging, some kind of event to then run the steam upload scripts at the right time

[tardy fossil]

does building through commandline with RunUAT.bat block the commandline? cause if it does you could make a bat file that runs packaging through the command line and runs steamcmd after

[grizzled stirrup]

Thanks I will try that

[meager spade]

you can mat a bat file to do all that

we have one

[hollow swallow]

Trying to just spawn 1 spell for each player. In my Actor Component (attached to player already) im running this. It spawns once for the host, but twice for the client. What am i doing wrong with my workflow?

[meager spade]

setlocal

:: Config
set UNREAL_ENGINE_DIR=D:\Program Files\EpicGames\UE_5.5
set PROJECT_PATH=D:\Projects\MyGame\MyGame.uproject
set TARGET_PLATFORM=Win64
set TARGET_CONFIGURATION=Shipping
set ARCHIVE_DIR=D:\Projects\TwinStick\BuildArchive
set STEAM_UPDATER_DIR=E:\SteamUpdater

:: Build + Cook + Package
call %UNREAL_ENGINE_DIR%\Engine\Build\BatchFiles\RunUAT.bat BuildCookRun ^
 -project=%PROJECT_PATH% ^
 -nop4 -platform=%TARGET_PLATFORM% -clientconfig=%TARGET_CONFIGURATION% ^
 -iterate -crashreporter -cook -stage -archive -package -pak -prereqs -distribution -build ^
 -target=TwinStick -utf8output ^
 -archivedirectory=%ARCHIVE_DIR%

:: Copy to SteamUpdater
cd /D %STEAM_UPDATER_DIR%

:: Clean previous content
if exist "%STEAM_UPDATER_DIR%\Content\" (
    rmdir /Q /S "%STEAM_UPDATER_DIR%\Content\"
)
mkdir "%STEAM_UPDATER_DIR%\Content\"

echo Copying to SteamUpdater...
robocopy %ARCHIVE_DIR% "%STEAM_UPDATER_DIR%\Content\" /COPYALL /E

call "%STEAM_UPDATER_DIR%\UploadGame.bat"

endlocal``` @grizzled stirrup

took out a lot of stuff we had

hopefully it gives you ideas

@hollow swallowofc it woul;d

host is spawning it for client also

so clients get 2

ran on server and client

so server will spawn it and so will client

[hollow swallow]

ah right, so do i just do a is locally controlled at the start?

so only the players version call to spawn it and not the server too?

or am i way off xD

[meager spade]

sure or just have the server create it?

and remove the rpc?

unless the beginplay is just for testing

literally just but a Has Authority

remove the server rpc

and voila

(assuming the spells are replicated actors)

[hollow swallow]

sweet that works. They are yeah

that spell will be creating another actor (projectile) but i only want local clients to create all of them, not have them replicated. For that when a player "casts spell" do i just do a multicast with the info, and each client will do the visuals themself? even for other players casting etc...

just trying to wrap my head around it all

[grizzled stirrup]

Thank you so much this is very helpful

No more waiting around to copy paste builds 😄

[grizzled stirrup]

Yes an unreliable multicast is fine for cases like that

[hollow swallow]

last thing before im off, when a client line traces and hits something, i spawn the projectile (only want clients to spawn it locally) it isnt replicated, and i pass through everything. When it hits i want the projectile to tell the server it has hit, and to deal damage. But the player reference is always invalid, even through i check its valid before it goes to the server call.

[grizzled stirrup]

Why are you passing an enemy reference when spawning the projectile? The reason your player reference is invalid is likely how you are setting it via the spawn projectile function can you screenshot that part?

[hollow swallow]

im passing it on so the projectile movement can follow its location, and also to then tell the server once its actually hit it. It seems to follow the location though with homing turned on, so it must be somehow valid

[grizzled stirrup]

In order to figure out why the player is invalid you'll need to log at each step the Player reference is used and see where it becomes null

Likely not getting set somewhere

[hollow swallow]

just before that section i check if its valid before even doing the linetrace

ooo actually i may have doubled my variables and used bp player instead of player xD

[grizzled stirrup]

right so it's valid when calling the server RPC but it's null in the projectile, how are you setting the player reference on the projectile?

[hollow swallow]

yep my bad, i doubled a variable and didnt notice. feel dumb now xD

[meager spade]

what is the iris

replacement for IsNetRelevent?

[quasi tide]

Group Filters I believe?

siliex shared some info with cedric a little bit ago. Touched on relevency #1363850238376415366 message

[nova wasp]

Filters and or net groups

[meager spade]

basically we have this ``` if (bWaitForReplication)
{
if (const AXX* WorldMapPlayerController = Cast<AXX>(RealViewer))
{
bool bGenerationFinished = WorldMapPlayerController->GetWorldGenerationLevel() == EWorldGenerationLevel::Finished;
return bGenerationFinished ? Super::IsNetRelevantFor(RealViewer, ViewTarget, SrcLocation) : false;
}

    // Should never hit this while using wait for replication
    ensure(false);
    return false;
}```

thats all we have lol

[simple remnant]

i switched it over but is auto possess player supposed to at player 0 or disabled?

[meager spade]

did they really make

NetRefHandleManager header private?

sooo i cant ever call it?

fk sake

[candid star]

what do you need it for?

the bridge should be able to expose some things to help

[nocturne quail]

Finally completed to design the inventory system that few days ago i was talking about, now its time to code it

I think 2k lines will cover all this stuff

[grizzled stirrup]

Does the amount of lines matter?

I just make it as clean as possible while doing the thing as designed

[nocturne quail]

it depends and if done properly by applying high level c++ techniques i think it will be super cleaner and less code

[meager spade]

what bridge?

[candid star]

the iris replication bridge

you still dodn't describe what you actually are trying to do though

[nocturne quail]

I mostly use templated parameters and avoiding Boilerplates

also one of my best technique to get the same result while keeping less code by using lambds

and local timers when it come to unreal where the timer no need to live longer or just a single call

this way no need to declare a new function to be called by timer

[grizzled stirrup]

local timer being the FTimerHandle is defined in a func instead of the header? wouldn't both require a new func to be called?

[nocturne quail]

yeah and use lambda along with it

[nocturne quail]

FTimerHandle UnusedLocalHandle;
GetWorld()->GetTimerManager().SetTimer
(
    UnusedLocalHandle,
    [this]()
    {
        //Code
    },
    1.0f, false
);

code will be called after 1sec

[grizzled stirrup]

Interesting I haven't seen that syntax before

[meager spade]

i showed the code

[nocturne quail]

C++ is magic

[meager spade]

in iris filters you only have nethandle

[grizzled stirrup]

curious what happens if you call this func twice in the same frame before the timer completes? Will it call // Code twice since the handle is local?

[meager spade]

i need to access the controller, but you cant

[meager spade]

icas cause most normal people dont write it like that :p

[grizzled stirrup]

😄

[nocturne quail]

nothing, i called like few thousand times few days ago

[grizzled stirrup]

I mean what's the result, does it run twice or only run once?

I assume it runs twice like a local timer would

[nocturne quail]

timer inside inside you mean?

FTimerHandle UnusedLocalHandle;
GetWorld()->GetTimerManager().SetTimer
(
    UnusedLocalHandle,
    [this]()
    {
        FTimerHandle UnusedLocalHandle;
        GetWorld()->GetTimerManager().SetTimer
        (
            UnusedLocalHandle,
            [this]()
            {
                FTimerHandle UnusedLocalHandle;
                GetWorld()->GetTimerManager().SetTimer
                (
                    UnusedLocalHandle,
                    [this]()
                    {
                        //Code
                    },
                    1.0f, false
                );
            },
            1.0f, false
        );
    },
    1.0f, false
);

here every timer is setting a new timer

and it works perfectly 😄

you can even set it to true to loop it

just dont to redeclare the timers again

i jsut copy pasted it

[nocturne quail]

most of them afraid of writing code in my style, that's why they have extra unnecessary defined functions for timers

not talking about those who don't know

Lambda is part of C++ and should be used

https://tenor.com/view/gigachad-gigacat-giga-gigabute-spyder228-gif-25745479

[candid star]

FReplicationSystemUtil::GetActorReplicationBridge

then bridge->GetReplicatedObject

[meager spade]

ah

ok

but it requires an actor

or UNetConnection

[candid star]

the filter should pass in the context to find the replication system

[meager spade]

no

all i have is ```cpp
struct FNetObjectFilteringParams
{
/**
* The contents of OutAllowedObjects is undefined when passed to Filter(). The filter is responsible
* for setting and clearing bits for objects that have this filter set, which is provided in the
* FilteredObjects member. It's safe to set or clear all bits in the bitarray as the callee will
* only care about bits which the filter is responsible for.
*/
UE::Net::FNetBitArrayView OutAllowedObjects;

/** FilteringInfos for all objects. Index using the set bit indices in FilteredObjects. */
TArrayView<const FNetObjectFilteringInfo> FilteringInfos;

/** State buffers for all objects. Index using the set bit indices in FilteredObjects. */
const UE::Net::TNetChunkedArray<uint8*>* StateBuffers = nullptr;

/** ID of the connection that the filtering applies to. */
uint32 ConnectionId = 0;

/** The view associated with the connection and its sub-connections that objects are filtered for. */
UE::Net::FReplicationView View;

};```

[candid star]

the view literally just has the player controller in it

[meager spade]

struct FReplicationView
{
    struct FView
    {
        /** The controlling net object associated with this view, typically a player controller. */
        FNetHandle Controller;
        /** The actor that is being directly viewed, usually a pawn. */
        FNetHandle ViewTarget;
        /** Where the viewer is looking from */
        FVector Pos = FVector::ZeroVector;
        /** Direction the viewer is looking */
        FVector Dir = FVector::ForwardVector;
        /** The field of view */
        float FoVRadians = UE_HALF_PI;
    };

    TArray<FView, TInlineAllocator<UE_IRIS_INLINE_VIEWS_PER_CONNECTION>> Views;
};```

?

[candid star]

let me see... I think the connection id should work

[candid star]

which filter class is this

most of them store the replication system as a member on init

which you can obtain the bridge from

::OnInit( passes in the replication system

so just add your own that does that

[meager spade]

yes

does what?

oh