archived-networking

[twin grove]

When a player connects to the lobby this should be enabled?

[twin grove]

Help guys

[tame slate]

Those buttons only show if the NetworkManager is not connected to a server.

[twin grove]

The debug log is triggered when connecting to the lobby, but StartClient(currentLobby.Value.Owner.Id); is not.

none of the logs are working here.

[tame slate]

That would imply that IsHost is true.

[twin grove]

I don't get it

like the player is the host?

StartHost work

[tame slate]

The only way that those logs don't fire is if the function returns because IsHost is true.

unless there's an error.

[twin grove]

Thanks, I figured out what the problem was, I had the transport hanging on another object:D

I forgot about it and used the getcomponent to get it.

[twin grove]

Now I'm faced with the fact that I need to create characters, although I have not even started the game, and I am at the lobby stage.

[sharp axle]

Do you need to create the characters in the lobby? Or are the characters being automatically created? You can remove the player prefab from the network manager if you don't want them automatically created

[twin grove]

okay, ty bro

then I can spawn them when the game starts?

[sharp axle]

Yes you can spawn them manually
https://docs-multiplayer.unity3d.com/netcode/current/basics/playerobjects/#spawning-playerobjects

[burnt ridge]

Hi all, I'm having an annoying issue with Unity Multiplayer Play Mode. "Pretty often" like once every 10 times, when I start MPPM with 2 clients (host + 1), the client will raise errors like:

The referenced script (Unknown) on this Behaviour is missing!

The host is fine. When it does that, I just kill it and restart it, and then it works for both host and client.
It's like it's not finished syncing the scripts or whatever for both the clients, but it sometimes happens even though I haven't made any changes between 2 starts.

Is this a known issue? Is there a fix?

[sharp axle]

What version of Unity and MPPM are you using? I've not seen missing scripts in the clones.

[burnt ridge]

6000.0.43f1 / 1.3.3

[sharp axle]

Mppm 1.6 just released. Might try updating

[burnt ridge]

Did it? can't see anything above 1.3.3 in packags

Maybe I need to update unity. Game releases in 2 weeks I'll pass for now ^^

[twin grove]

What can cause me to get a kicking error? But the player seems to get kicked normally.

[sharp axle]

That is just a warning. Looks like it might be trying to remove the player twice for some reason.

[twin grove]

So there's a problem here?

[sharp axle]

its not really a problem. just the host there is already disconnecting the client. The client RPC is calling SteamLobbyManager I don't know what it does, but might also be trying to disconnect the player

[twin grove]

SteamManager

[sharp axle]

Ok, so the host has already disconnected the client, you don't have to call Shutdown again

[twin grove]

ty i delete NetworkManager.Singleton.DisconnectClient(targetClientId);

[bitter vigil]

hey hi! im kinda new to unity, i got some expirience at this point but not enough for netcode for gameobjects, so was curious, is it a bad idea to code stuff first and work in the networking later? and if it isn't a bad idea, how hard would it be to do it now and to do it later? 👀
I am mostly asking this out of curiosity so pls dont hate me for it!!

[elfin field]

KeyNotFoundException: The given key 'ChatManager' was not present in the dictionary.
System.Collections.Generic.Dictionary`2[TKey,TValue].get_Item (TKey key)

[runic pivot]

This is kinda... the only way to learn it, in my opinion, but I ended up with so much clutter when I did it. Not knowing how things were going to work on the server side means I designed a lot of my systems poorly, and when I was duplicating code to move functions from client to server side, I ended up with a lot of mess and functions that were only half useful after the fact. If you're just thinking about implementing multiplayer later and won't be moving game systems to cloud code modules, it might not be too bad

[bitter vigil]

so basically learn a lot more unity and then later decide what server needs to process and what client needs to do and change the stuff needed?

[sharp axle]

It really needs to be designed for networking from the start. unless you don't mind completely rewriting whole chunks of your systems

[bitter vigil]

hmm yea fair, the thing is, i have some cool ideas for a game but it also kinda needs to be multiplayer to be fun, i currently have the ideas and motivation which kinda makes me itch to start, which is why i asked :D
how much of a difference is there in code of normal unity and netcode if i may ask?

[runic pivot]

there's just more to know. Learning unity and multiplayer things at the same time is gonna be hard, but if you're dead set on this project, give it your best.

but if you're somewhat comfortable in unity already, then I would start with learning the netcode side. There are some youtube tutorials that might help

[naive wave]

Hey all. I'm new to Unity as of about a month ago and I got a fully functioning single player game going. I added some netcode and tested locally and now I'm to the point where I tried to integrate dedicated server support via Unity's Matchmaking + Multiplay Hosting.

My matchmaking works like a charm. I get my response ticket with the IP and Port of the session. My server starts up properly according to my server logging and is waiting for my clients to connect. Then I try to connect with my two clients and..... nothing. Failed to join server. I'm using the IP and Port from the matchmaking response ticket when trying to start the clients. Any ideas what I might be missing here?

[sharp axle]

The code itself is not really any different. Its more about designing your systems to be more data driven so things can be sent across the network. Mapping out which clients will own what objects and whatnot

[bitter vigil]

ohh fair fair! thanks for making that clear btw!

[bitter vigil]

hmm, i think im gonna learn a bit more 3d and just code in general and then attempt it, i feel like i need to make some normal stuff first, like i have done some 2d but not that i feel like im standing too steady with it yet, so imma practice it first and then attempt to edit the code/rewrite it!

[sharp axle]

From a cold start, the server might take a minute or two to fully spin up. But really depends on what error message you are getting

[bitter vigil]

another curiosity i had is how it works when one makes a game with single player and multiplayer like mc

[runic pivot]

mc = minecraft?

[naive wave]

The server isn't giving any error message and the client is timing out after trying to connect for 10 seconds. And then retrying and the same thing happens again.

The sever seems to be fully spun up according to logs.

[bitter vigil]

yep

[naive wave]

@sharp axle Is it cool if I ask you a few things in a private chat?

I might be duplicating some logic that happens automatically in multiplay, unclear.

[runic pivot]

alternatively, you could make a thread here

[naive wave]

oh sweet

Didn't know that was a thing

found it, ty, making one!

sorry taking me a while to make it, adding some additional questions in there.

[sharp axle]

The easiest way would be for single player to be the same as multiplayer just connecting to localhost.

[naive wave]

Unable to connect to Server (Matchmaker + Multiplay Hosting dedicated server)

[naive wave]

I second this. You can do that first and add server support later. That's what I'm in the process of doing.

[bitter vigil]

so just setting the server limit to 1?

[naive wave]

Nah, your computer is the server basically, the computer you are running the code on.

[sharp axle]

no, it just connects to itself as host

[bitter vigil]

mhm ik, just host and having no-one connect

[naive wave]

you can emulate the connections using multiplayer services, you can have like two players connect in different windows, it's pretty nifty

They even have the console in each window so you can see all of your debugging stuff right there.

[naive wave]

Anyone know why when "using Unity.Services.Multiplayer" I can't create an object of type "IMultiplaySessionManager" ?

[sharp axle]

It can only be used on the server so the build platform has to be set to dedicated server.

Or you can add UNITY_SERVER to the compiler defines in the project settings

[twin grove]

all the players are ready, how do I start the game with the players?

[sharp axle]

The host needs to switch scenes with NetworkManager.SceneManager.LoadScene the other clients will automatically sync up

[twin grove]

okay, ty bro

[twin grove]

What if the host quits during the game? How do I get the player back in the lobby menu?

[sharp axle]

If the host quits, then all clients will get disconnected. They will fire the OnClientDisconnected callback or the ClientDisconnected Connection Event. When that fires you can use the regular Scene Manager to go back to the menu scene

[twin grove]

it's not working for me.

[sharp axle]

What part is not working? None of the code you posted is changing scenes

[twin grove]

Scene change works and the players are thrown, but if the host closes the game, the players are not thrown out of the game in the menu.

[naive wave]

Hm... anyone ever see an issue where your CPU on your dedicated server spikes when the game initializes? This is a 2d 1v1 cardgame with almost no graphics currently.

[delicate parrot]

Can i make a general struct implementing inetworkserializable?

eg.

    public struct NetworkValueWithRarity<T> : INetworkSerializable where T : struct, INetworkSerializable
    {
        public T Value;
        public int Rarity;

        public NetworkValueWithRarity(T value, int rarity)
        {
            Value = value;
            Rarity = rarity;
        }

        public void NetworkSerialize<T1>(BufferSerializer<T1> serializer) where T1 : IReaderWriter
        {
            serializer.SerializeValue(ref Rarity);
            Value.NetworkSerialize(serializer);
        }
    }

im worried it won't like the lack of definition

[green vortex]

im now implementing privacy settings , where users can change their session visilibility by changing 2 variables + 2 properties inside ISession

but i cant

it turns out that the session is still didnt change

heres how i query the session

-> option:
SessionFilterOption = new QuerySessionsOptions()
{
    Count = option_maxQuery
};

    public async Task UpdateSessions()
    {
        QuerySessionsResults result = await MultiplayerService.Instance.QuerySessionsAsync(SessionFilterOption);
        AllActiveSessions.Clear();
        AllActiveSessions = new(result.Sessions);
    }```

heres how i change it

public static void SetPrivacyState(this ISession session, SessionConstants.SessionPrivacy privacy, string password = null)
{
    IHostSession sessionConfig = session.AsHost();

    sessionConfig.IsLocked = privacy is SessionConstants.SessionPrivacy.SinglePlayer;

    sessionConfig.IsPrivate = privacy is SessionConstants.SessionPrivacy.SinglePlayer;

    sessionConfig.UpdateFriendOnlyStatus(privacy is SessionConstants.SessionPrivacy.Public_friend or SessionConstants.SessionPrivacy.AskToJoin_friend);

    sessionConfig.UpdateAskToJoinStatus(privacy is SessionConstants.SessionPrivacy.AskToJoin or SessionConstants.SessionPrivacy.AskToJoin_friend);

    if (privacy is SessionConstants.SessionPrivacy.Private)
    {
        sessionConfig.Password = password;
    }
    else
    {
        sessionConfig.Password = String.Empty;
    }
}```

the default privacy option is public

lets say host A changed his room using this function and then B queried, what B saw is still public

[green vortex]

the latest session returned by QuerySessionsAsync will ignore changes u done in ihostsession(or session.asHost())???

[sharp axle]

After you change session properties, make sure you calling IHostSession.SavePropertiesAsync()
https://docs.unity.com/ugs/en-us/manual/mps-sdk/manual/session-op-host#write-apis

[green vortex]

ohhhh

ty👍

[potent monolith]

Hi, I'm confused as to how to retrieve a JSON from the server in OnNetworkSpawn for my small game
I'm making a game for just a few friends to embed as a discord activity so I'm storing files on the server as JSONs

I'm ab le to serialise the JSON to and from a PlayerData class, so I thought I would have the client request the JSON string from the server with a ServerRPC

And then OnNetworkSpawn would look like

void LoadData()
{
  PlayerData data = PersistenceManager.LoadData(id);
  // ...
}

public override void OnNetworkSpawn()
{
  LoadData()
}

And PersistenceManager would look like

public static PlayerData LoadPlayer(int id)
{
  string json = instance.LoadJsonFromServerRpc(id);
  // (deserialise)
}

[ServerRpc]
string LoadJsonFromServerRpc(int playerID)
{
        string path = Path.Combine(Application.persistentDataPath, playerID.ToString() + ".json");
        if (!File.Exists(path))
        {
            File.Copy(Path.Combine(Application.persistentDataPath, "default.json"), path);
        }

        return File.ReadAllText(path);
}

However, I'm not allowed to return values from a server RPC. Online it says I should have the server call a clientRPC, but I don't get how I would pause during OnNetworkSpawn until the players data has been loaded

Also not sure how to link the player object that is requesting the loaded data to the server incase the server has to call a ClientRPC

[sharp axle]

You can't pause to wait for an RPC. The easiest way would be to use a NetworkVariable<FixedString> to pass the JSON to the clients. Alternatively you can use the OnNetworkSpawn() on the server to send the string in an RPC

[potent monolith]

If I do the second one what would the sort of method flowchart look like

Like, how would it ensure the players data is loaded before they get control

And would the RPC be on the Persistencemanager object? If so would it be

Player sends reference to persistencemanager game object in onnetcodespawn
Persistencemanager does serverrpc to get the string
Server invokes a clientrpc on PersistenceManager to send the string to the player?

Sorry for capitalisation I'm on phone rn

[sharp axle]

You can do it this way. I don't know what all PersistenceManager is doing or what the client is doing with the JSON.

If all clients need the json then you could different things than if only the one client needs it

[potent monolith]

Essentially the JSONs are on the server's filesystema nd the PersistenceManager has the (static) ability to deserialisie a JSON to a Unity object

[potent monolith]

Every client would need their own JSON

So the server needs to know which client asked for it

I thought this could be done by setting a reference to the player gameobject to the PersistenceManager

my problem is not being able to await the JSON

If I set a NetworkVariable<FixedString>, what if OnNetworkSpawn finishes before the JSON is available?

And if I do an RPC, I'm not sure what to do if it isn't called before the player does an action like openiing their inventory

I'm also considering doing a POST request to get and save JSONs and host an endpoint for that but I'm not sure if thats a good idea

[woven onyx]

I want to create simple synchronized interactions for the same interactable object across multiple clients. All clients should be able to interact with it, and these interactions must be synchronized over an intranet (local network), without using the internet. How to Achive this ?

[sharp axle]

If you use the OnNetworkSpawn() on the playerobject you can pass the player's ownerclient Id to the PersistenceManager. it can then send an client RPC to that specific client.

[sharp axle]

You can use RPCs over a LAN connection

[woven onyx]

is Unity Transport (UTP) will be helpfull?

[sharp axle]

You dont need to mess with UTP directly. If you know the IP address you can either enter it directly or use something like Network Discovery to find the host IP

[woven onyx]

@sharp axle Thank you

[potent monolith]

Right, I'll use that pattern then
Thanks

[formal atlas]

I need help setting up online multiplayer. I am using the VRMP template on Unity 6, and I cannot figure it out. (I test with the Multiplayer Center) The rooms show up on both players, but the players or interacting does not.\

[sharp axle]

It sounds like the clients are not joining the same session

[heavy ferry]

Hi Guys,
Does unitywebrequest work like coroutine and on main thread? I have to download 6-7 glbs . Not sure if i should use c# native httpClient or unitywebrequest.
I heard UnityWebRequest is executed mainly on main thread.
Dont want to make the game stutter during the downloading (i dont trust coroutine based behaviour for performance while downloading large files paralley)
If i can offload the download to other threads then all good.

[potent monolith]

Does anyone know if it's possible to do animation blending for Network Animators? I have a client authoritative network animator on my players and animation blending only occurs for the users player, other players in the session play the right animations but don't blend between them. Is there a simple and modular way to synchronise the other parameters or should I manually write a script to synchronise animator parameters as network variables on updates?

[sharp axle]

check the example

[sharp axle]

As far as I know, Animations should still be blending on remote players. There might be issues with animation events though

[heavy ferry]

Isnt this working on coroutine?
If i have many models then wont it hog the main thread and bring down the FPS

[tame slate]

Why are you loading glb files specifically during gameplay?

[potent monolith]

Probably an events thing, i have events for the things I was concerned abt (attacking and jumping)

Im ok with it for players

but is there a better way to avoiid this for server authoritative enemies?

[heavy ferry]

Just trying to load some objects before game starts and less prio items during gameplay in background

[fluid walrus]

the actual implementation of web requests is native and platform specific, i doubt most platforms do the heavy lifting on the main thread

[sharp axle]

If its just waiting on downloads its not gonna hurt the framrate.

[tame slate]

Yeah that should be completely fine when done asynchronously. You might find Addressables easier to use. Can convert those glb files in to ready to use prefabs that can be fetched remotely at runtime.

[twin grove]

is there any way to get a real ping to the host? Facepunch

I'm getting fake ping right now.

[tame slate]

https://docs.unity3d.com/Packages/com.unity.netcode.gameobjects@2.4/api/Unity.Netcode.NetworkTransport.html#Unity_Netcode_NetworkTransport_GetCurrentRtt_System_UInt64_

I think that should work as it's from a generic class

[sharp axle]

I don't think the Steam Transports implement GetCurrentRTT()

The next best way would be to look into the PingTool

[twin grove]

how do I download it?

[sharp axle]

Easiest way is probably to clone the repo, then copy the PingTool folder from the Examples

[twin grove]

you can't get through this?

[sharp axle]

not with FacePunch

[twin grove]

😦

I don't see any downloads here.

[tame slate]

You’d have to clone with Git or GitHub. Or there are also websites that will download repos for you.

You could try to add it as a package through package manager but I’m not sure if or how that’ll work for this specific repo.

[twin grove]

it won't work with Facepunch?

[sharp axle]

Nope. I just showed where the the Facepunch Transport will always return 0 from that. There doesn't seem to be a way to get ping through Steamworks.

[twin grove]

It's sad, but will what you're throwing in there work?

[sharp axle]

The PingTool? yea. its using RPCs to get the ping times.

I guess there is this but no clue how or if it works

[twin grove]

I only downloaded the tools folder.

did you use it yourself?

[sharp axle]

I've used it before. if you downloaded the whole example, it would have included the Network Stats Monitor

[tame slate]

It looks like there’s direct references to UnityTransport in there. I don’t think that’ll work for Facepunch either. Not sure how much of a hassle it would be to adapt it.

[sharp axle]

It will list both the GetCurrentRTT() value and the ping through the RPCs. the RTT value would be 0 but ping values would still be valid

[twin grove]

https://wiki.facepunch.com/steamworks/SteamNetworkingUtils.EstimatePingTo

[twin grove]

I checked with another dude who lives far away, still shows a crooked ping.

[ashen meadow]

Hi everyone, I'm making a chat system right now just to get familiar with coding multiplayer, and I was wondering what the best way would be to make sure an RPC is coming from the server

[tame slate]

What library are you using?

[ashen meadow]

Just netcode for gameobjects right now

[tame slate]

All RPCs sent to clients are technically from the server.

You can explain further if you're trying to do something specific, but if you're using the new Rpc attribute to do client to client RPCs you can put

if (!IsServer) return;

at the top of the function to ensure that only the host/server can send the RPC.

[ashen meadow]

@tame slate I was more concerned with cheaters but I don't know enough about networking to even know if that's a real concern

Right now i was just checking if the client id was 0 to see if it was from the server but I have no idea if that was the right idea

[tame slate]

@ashen meadow you would use this in the SendChatToEveryone function, not the SendToServer one. The way you have it now would just prevent anyone from sending their chat to the server.

But regardless of all that, I wouldn't worry about any of these checks for what you're doing currently.

[ashen meadow]

Are you sure? The chat is working right now

[tame slate]

Only the host will be able to send messages.

[ashen meadow]

My clients can too though?

[tame slate]

If they call SendToEveryone directly, sure. I meant if they were using SendToServer in order for the chat to be sent to everyone. I am not sure of the specifics of how you set things up.

[ashen meadow]

This is my update function where I read in the chat from the input field

It's a host and clients setup

Here's the whole thing hopefully that makes it easier to read

[tame slate]

You have two references of SendChatToEveryoneRpc. Where is the second reference?

[ashen meadow]

Just the intro message

[tame slate]

Huh. Yeah any function with if (!IsServer) return; should not be able to run on a client. But anyway, this is kind of irrelevant because you shouldn't need to use that check for what you're doing. What you have is fine without it.

I think you're on the right track with checking the sender ID to make sure it's from the server before displaying it, but in all honesty, if someone is already decompiling your game and cheating, they could get around that quite easily, so I wouldn't really worry about it.

And I also wouldn't put too much weight on focusing on anti-cheat measures as a beginner. It can be an extremely complex topic and is only really valuable for production level games.

[ashen meadow]

Yeah I was more just interested, thank you for all the talk though it makes me feel better about what I've currently got

I'm pretty happy with just checking the sender client id at least

It'll probably be like a 4 player steam friends only game anyways if it goes anywhere

Tyvm!

[haughty river]

i was wondering if anyone could tell me how to make a queue up button so people coul 1v1 in a arena because i have searched for tutorials and none is giving me what im looking for

[jolly tinsel]

hi there, quick theoretical question. if i were to use a raspberry pi (say pi 3 or older because of cost and availability) as a server for a game, what would be the limit of informations i can handle? like, is real-time-movement without too many delay possible (assuming i have a stable enough internet connection) or is the raspi just too slow
thank you 🙂

[bright coral]

If it's just about relaying data, the pi can probably handle quite a bit. Most of your resource utilization will come from simulations and other server side functions. But just syncing data? Probably it can handle quite a few players.

[jolly tinsel]

so, if i just want to connect 2 players positions and rotations, that would be doable right?

if i do the rest client side, ignoring obvious security risks 😄

[bright coral]

Sure, but it also depends on how you write your server. Running a Unity instance as a server will be a lot more resource intensive than having a barebones server written in Go for instance

[jolly tinsel]

ok, can you recommend some ressources for me to research?

[bright coral]

On what specifically? I sense maybe you're using NGO or some other framework that expects you to build a server instance from Unity?

If you just want to do very simple network syncing, I can highly recommend https://github.com/RevenantX/LiteNetLib
Easy to implement in Unity for your client, and for the server you just create a barebones .net console app.

[jolly tinsel]

alright, thanks, ill look into it 🙂

[bright coral]

Here's one of my favorite resources if you're looking into multiplayer networking more broadly
https://github.com/0xFA11/MultiplayerNetworkingResources

[jolly tinsel]

damn, thats a lot (to be expected :D)
thank you very much, this ll probably take some time 🙂

[hybrid dew]

I can't for the life of me think of a clean way where a game can start as singleplayer and then the player create host and allow others to join them

Do I just design everything as if it's a multiplayer game
Just instantiate while it's singleplayer
and when player chooses to host, I just loop over everything in the scene, and call Spawn on them with a nullable check on Network Object?

seems dirty af

[bright coral]

This might be helpful
https://discussions.unity.com/t/does-netcode-for-gameobjects-have-offline-mode/863780
They talk about switching transports at runtime. Essentially your singleplayer mode would be a local session where your player is the host, and later on if you want to open it up you switch the transport to whichever one you use for networked sessions.

[sharp axle]

Yea, the easiest way to do single player would be start the game as host connecting to localhost. All spawn logic would remain the same. A new Single Player Transport was recently added to the latest NGO 2.0 version.

[haughty river]

im having this problem where it says the player prefab has spawned in the debug logs and everything but nothing is appearing like no prefab in the hiearchy or anywhere

[sharp axle]

If you are not seeing the prefab in the hierarchy then its not getting spawned. Are you spawning it manually? or are letting the network manager spawn it automatically?

[haughty river]

at first i let the network manager spawn it automaticlly but that didnt work so then i tried manually and that still didnt work

[sharp axle]

There is no reason for the automatic player spawn to not work. Unless the prefab is disabled.

or the network manager isn't actually connecting

[haughty river]

idk whats happening then

[sharp axle]

throw a network object on a cube and see if the same thing happens

[haughty river]

ok

no the cube clones itself

i just switched out the cube on the list in the default player prefab

so i just tested some stuff out and if i set it all up with the cubes everything works fine but for some reason it wont work with my model

[upbeat summit]

[ServerRpc(RequireOwnership = false)]
public void RegisterToolServerRpc(int index)
{
    if (index < 0 || index >= toolsList.Count) return;
    ToolInfo toolInfo = toolsList[index];
    ToolInfo.ToolData toolData = toolInfo.tool;
    GameObject toolPrefab = toolData.toolPrefab;
    if (IsServer && !registeredTools.ContainsKey(index))
    {
        // SERVER: Spawn tool instance
        GameObject toolInstance = Instantiate(toolPrefab);
        NetworkObject toolNetworkObject = toolInstance.GetComponent<NetworkObject>();
        //hope this is fine, i think i need this check to prevent warnings
        if (!toolNetworkObject.IsSpawned) 
        {
            toolNetworkObject.SpawnWithOwnership(OwnerClientId);
            Debug.Log("Spawned tool "+toolPrefab+" with ownership "+OwnerClientId);
        }
        else
        {
            //Debug.LogWarning($"Tool {toolInstance.name} already spawned!");
        }
        // Save offset data for later placement
        toolInfo = toolInstance.GetComponent<ToolInfo>();
        toolData = toolInfo.tool;
        registeredTools[index] = toolInstance;
        toolInstance.SetActive(false);

        //Debug.Log($"[SERVER] Spawned and registered tool: {toolInstance.name}");
    }
    else
    {
        // CLIENT: Wait for server-spawned object
        //StartCoroutine(WaitForToolSpawn(toolPrefab.name, index));
        NetworkObject netObj = registeredTools[index].GetComponent<NetworkObject>();
        ulong netId = netObj.NetworkObjectId;
        SyncToolToClientRpc(netId, index);
    }
}

[ClientRpc]
private void SyncToolToClientRpc(ulong toolNetworkId, int index)
{
    ToolInfo toolInfo = toolsList[index];
    ToolInfo.ToolData toolData = toolInfo.tool;
    GameObject toolInstance = null;
    //TOOL LOOKUP
    if (NetworkManager.Singleton.SpawnManager.SpawnedObjects.TryGetValue(toolNetworkId, out NetworkObject netObj))
    {
        toolInstance = netObj.gameObject;
    }
    else
    {
        Debug.LogWarning($"No NetworkObject found with ID {toolNetworkId}");
    }

    Debug.Log("Registered gun "+toolInstance.name+" with owner id "+netObj.OwnerClientId);
    // Check if this is the one assigned to this player
    if (netObj.OwnerClientId == OwnerClientId)
    {
        registeredTools[index] = toolInstance;
        toolInstance.SetActive(false);
        ToolInfo offsetToolInfo = toolData.toolPrefab.GetComponent<ToolInfo>();
        ToolInfo.ToolData offsetToolData = toolInfo.tool;
        // Copy the transform data from the prefab directly, because not doing this was very stupid!!!
        toolData.posOffset = offsetToolData.posOffset;
        toolData.rotOffset = offsetToolData.rotOffset; 
        toolData.ADSposOffset = offsetToolData.ADSposOffset;
        toolData.ADSrotOffset = offsetToolData.ADSrotOffset;
        toolData.toolWeight = toolInstance.GetComponent<Rigidbody>().mass;
        Gun gunScript = toolInstance.GetComponent<Gun>();
        gunScript.OnToolEquipped();
        // Update the tool's position and rotation like in the server code
        toolInstance.transform.position = transform.position + transform.rotation * toolData.posOffset;
        toolInstance.transform.rotation = transform.rotation * Quaternion.Euler(toolData.rotOffset);
    }
}

Hello, i am having an issue with networking and i would be very thankful to receive any help, heres the problem in detail

2. server then checks the validity of register tools, and also the client upon joining calls everyones register tools, the server checks if the already registered tool dictionary contains key the client claims it doesnt, and if it does, it will not spawn in a new gameobject, but instead tell the client about the existing one.
- For some inexplicable reason, the client ends up with tool duplicates, the duplicate tools cannot be seen server side and always spawn at 0,0,0 and mess with the game because they can infact shoot.

Nowhere else in the code is “instantiate” or “spawn” called

[potent monolith]

followup for this, but I have ideas to sync animator parameters now for blending stuff like walking animations based on movement speed: but I'm wondering if anyone knows how to solve the issue of Triggers not blending
For example I have a transition from Idle to Jump with a trigger, and the blending only works for the owner of the object
I'm curious about having server authoritative enemies who do things like attack based on a trigger and how to synchronise their animations to blend smoothly for clients

[twin grove]

what are these errors related to?

[sharp axle]

I would use NetworkObjectReference here. Then on the server reparent the network object after you spawn it. I believe as of the latest NGO version, the owner can reparent the object

[upbeat summit]

Thanks! I had no idea you could do this, i thought i had to manually search for the network object each time, its very cool than you can actually pass them into server/client rpcs!!!
Also i dont do parenting because the tools have their own networkobjects, so they cannot be children of other networkobjects, in this case the player.

[delicate parrot]

(behind the scenes it just passes over the unique hash on the spawned object because internally ngo keeps a dict of the serializable hash and the non serializable object)

it's pretty easy to make equivalent solutions for other non-serializable objects that will have identical dict or list indexing where you just wrap around that

[twin grove]

help bro

[sharp axle]

Network Objects can be reparented after spawning. You just have to call NetworkObject.TrySetParent after spawning

[upbeat summit]

your networkprefab asset has a null entry if i had to guess, which means theres a missing object

networkprefabs is a thingy in your folder

[sharp axle]

no clue where those network prefabs are. But unity thinks they are invalid for some reason

[upbeat summit]

double click on the networkprefabs thingy right there

[upbeat summit]

then click minus

[twin grove]

Thank you. Why is that? I just updated the engine.

[upbeat summit]

it is because there was a missing entry, with nothing inside of it, and unity netcode doesnt like that

you should only add entries when you have a new networkobject that you actually want to add to it

[twin grove]

It's just, it didn't happen before, it happened after the engine update.

[upbeat summit]

🤷‍♂️

[sharp axle]

I didnt think it would even let you add a object there without a network object attached

[twin grove]

Who knows how I get +- correct ping to the hoster I use Facepunch

[sharp axle]

You can use Ping/PongRPC . NetworkManager.LocalTime.Time - NetworkManager.ServerTime.Time should also give you a rough estimate

[twin grove]

Okay, thanks. I'll give it a try.

[silk bluff]

Iam currently trying to use the relay system from unity. I have the project connected and everything. My code rn will create the game and give me the code (i can see a player joining on my dashboard) but then when trying to find that connection with the code on another device i get "join code not found"

i tried rewriting my script and connecting everything again and no dice

https://paste.mod.gg/fameshdmfbbt/0

nvm i just found that the problem was i was trying to join a session from the same computer

[waxen quest]

Is there a way to change tick rate through script?

[sharp axle]

There is NetworkConfig.TickRate but I don't think you can change this after starting the server/host

[waxen quest]

Yes, I only want to change before starting the host. Thanks I'll use it

[tidal hemlock]

Hey not sure if this is actually a networking issue, but only happens on other clients so 🤷‍♂️

Some connected clients, when they are spawned, fall through the floor:

• ONLY HAPPENS SOMETIMES (hard to debug)
• Character Controller (no rigidbody)
• colliders are not set to trigger
• only happens on spawn, if they dont fall through floor on spawn the rest is fine
• custom/simple gravity logic for character controller but i dont think thats the issue

(The image is in paused state, but it just keeps falling)
Any help is appreciated

[tame slate]

Are you spawning players directly on the ground's Y level?

Usually helps to provide some buffer between the spawn point Y level and the ground

[tidal hemlock]

Yeah, have at the same level, but its weird that it works sometimes and not other, I'd rather not increase it so players are not falling (even if just a bit) when game loads, but if theres no other option 🤷‍♂️

[tame slate]

You could also try disabling your gravity logic until OnNetworkPostSpawn?

[tidal hemlock]

Didnt even know that was a method 🫠

will try that

[tidal hemlock]

did a little more digging, i was spawning my players in onNetworkSpawn, and sometimes on some devices scene takes a few frames longer to load than when onNetworkSpawn gets triggered so those players were falling through, moved the spawning to the scene loaded event and now its working 🤞

[old mural]

Is it better to run one game server (GS) process on one VM, or many GS processes on one VM?

[sharp axle]

Depends on how you are paying for your VMs

[old mural]

What do you mean?

[sharp axle]

It might cost you more money to have multiple VMs running or it might cost more to have a larger VM that can run multiple servers

[old mural]

Ic. Basically if it’s cheaper to scale horizontally, then 1 GS process per VM makes sense. If it’s cheaper to scale vertically, pack multiple GS processes on same VM. Any nuances to make note of?

[sharp axle]

Just whether the hosting provider can scale the servers automatically for you or not. And if/how you get charged for sudden usage spikes.

[old mural]

Seems like it’d be more commonly cheaper to pack one VM with many GS processes. Was checking some stats and depending on the game type, you can scale to hundreds or thousands of GS processes. I specifically asked for AWS resources but YMMV.

What hosting service(s) do you have experience with?

[sharp axle]

I've only used AWS, Multiplay, and Azure Playfab. You are not getting that many Unity servers running on a single VM

[old mural]

Why do you say that?

[sharp axle]

A Unity Dedicated Server uses a lot of resources, even stripped down to no graphics.

[old mural]

Reran calculations using 60%-70% utilization expectation and found that the more realistic estimation is from dozens of GS processes to the low hundreds of GS processes for common AWS instance types. That was helpful to think through 🫡

[haughty river]

i got a problem so the player prefab spawns when you are host or client but when i click both in the host hierachy only one of them shows but when i do it with cubes both of them show in the hiearchy

[haughty river]

its fixed lets go

[hybrid robin]

How i can made a central server for my game? Using mirror

[sharp axle]

You are going to have to be more specific.

[hybrid robin]

Main server

Do u know what is that?

[lyric thunder]

any way i can call PhotonNetwork.Instantiate without using room objects or the resources folder?

[green vortex]

lol nice we have our own section here

i use netcode + unity session

[sharp axle]

I know how to start as a server. But I don't know what you mean by "Main/Central" server

[dark umbra]

what underpaid intern was high on crack when writing this sentence?
Sometimes, you might want to make a simpler prefab instance to be spawned on server version the override for clients.

[robust lion]

i downloaded PUN2 but when i press open in unity i dont see the asset in the package manager

[sharp axle]

PUN2 is way out of date. You should probably be using Fusion 2 at this point

[robust lion]

ok

[torpid stream]

what

what you mean

[modern ginkgo]

moderator already remove post in question

[torpid stream]

😄

they are like robots

they never talk to us

[modern ginkgo]

[torpid stream]

Hello, I have been developing games for 6 years, but I have never tried implementing online multiplayer features before. I have a very promising idea in mind, and I need to publish it on Google Play as soon as possible. Could someone please share a detailed article or guide about using Unity Networking and implementing online mechanics? This game is extremely urgent, and it’s the best idea I’ve ever had in my life.

[tame slate]

There are endless ways to do networking in Unity and choosing how to do it almost entirely relies on what your game requires and your personal preference. You should be able to find a handful of popular solutions just by Googling or searching on YouTube.

[torpid stream]

I couldn't find any decent resources, what I'm looking for is to set up my own dedicated server.

I have a hosting company that I own completely, so I don't have any server problems. I just created a high-quality VPS Server for myself.

So how do I set up a server and connect to it in the game?

[tame slate]

That would be one of the later steps of doing multiplayer development and again depends on which solution you plan on implementing.

[torpid stream]

What do you mean by which solution?

here is my own solution my own vps server?

[tame slate]

I'm referring to how you would actually implement multiplayer networking in to your game. There are plenty of low-level and high-level frameworks/libraries you can use. They all have their own way of connecting to remote servers.

[torpid stream]

Which ones do you recommend, I have no knowledge of these?

What are the cheapest or completely free network connection solutions that would be sufficient for me?

[tame slate]

#archived-networking message

Most of them are free. People here commonly use Netcode for GameObjects, Mirror, Photon, FishNet and a few others.

[torpid stream]

i think netcode the best choice for me

thank you so much

[waxen quest]

Is there a way to detect collision on network objects

Like I have a network ball and I have to play sound for all clients when the ball hits other objects

The problem is it only detects collision on host and for other clients the ball does not properly collide because it is networked and it changes position early

I use Rpc as of now but looking for a better approach

[sharp axle]

When using Network Rigidbody, the host/server is the one that handles the physics. If you want physics to be handled locally, you can look into using Distributed Authority.

[midnight tartan]

hi, i develop an rpg and i have no clue at all about networking. whats the best to start with networking for mmorpg?

[waxen quest]

Ohh I'll check that out

[sharp axle]

First, you need to drop the massively multiplayer part. You are not making one without a large team and a huge budget.
Second, pick a network library and start some tutorials. I use Netcode for Gameobjects and recommend the CodeMonkey multiplayer course.

[midnight tartan]

i did the samyam multiplayer snake course. it worked. but i am not sure if that can handle many players and stuff. i did an offline arpg for 2 years so im good with offline games and programming stuff. we start as 4 people and i think its fine, nothing to rush. fishnet and netcode are kinda the same thing?

[sharp axle]

all the modern network libraries are broadly similar.

[spring walrus]

there is Mirror ummorpg which costs a lot. I doubt that any open source networking solutions can reliably handle that many connections at once. I`d look into separating big regions of the map into small ones and having each one a separate server, like Foxhole did, though this will skyrocket server cost.

I meant the modified Mirror of ummorpg, sorry if I worded unclearly

[midnight tartan]

understand, thank u sir.

[coral stratus]

Would it be better to handle things like bullet penetration/ricochet and tumbling on the server and replicate it off to clients? Or should I simulate this on the client and hope the server also simulates something identical?

[sharp axle]

That is called client prediction. As long as your bullet calculations are deterministic, client and server will have the same results. Ideally, you would handle Rollback on the clients in case of any mispredictions

[green vortex]

unity got way way way more choice u can think of , and honestly i do not recommend you to implement ur own backend from scratch unless ur solely learning or really know what to do

there are legal issues if you want to publish ur game in china

which is something i need to think of as well

one thing is all the packet/data transfer must not go outside of china, like u need to isolate ur game backend by china and places other than china

ur game will be flagged for security concern if u dont obey this rule

[sharp axle]

Thats a good point. I don't think any of the Unity Game Services have servers in China

[green vortex]

i think the most difficult part is ensuring all the network routing is fully inside china

like if the data transfer route has 0.1% thats outside china, ur game will be rejected

there are 3 more rules you need to obey tho

1. u must register/setup a legit company in china
2. ur game servers/database must be in china (you can have other servers, but need to have an isolated server in china) , the isolated chinese server must not able to exchange data with other servers
3. you must give players an option to delete their player data (character and other stuff) , the deletion must be a complete wipe , u need to ensure none of the data is left on ur server/cloud/database...etc

they are strict , because chinese players must register games with their ID numbers,address ...etc , they need to be identified

its required by law

as for why they do that, anti-addiction, to prevent children addicted to games

but these are goals after my game are out, no need to think of atm 🤣

lets hope i only need to swap out DGS but not netcode as well lol

[torpid stream]

I living in Turkey. Do you have any information about whether there is such a rule in Turkey?

[torpid stream]

😄

[green vortex]

why would i lol, i live in hong kong

welp all i can say is u need to do research urself

[torpid stream]

OMG! Is Hong Kong in China?

I thought it was in Thailand

[blissful jay]

🤨

[green vortex]

lol

[potent monolith]

Does NetworkObject.Despawn() not synchronise from server to client if the NetworkObject is in-scene placed?

I have a coroutine as an animation event as

public virtual void FinishDeath()
        {
            // Any death events happen here, then two couroutines to despawn the object and animate the death animation
            if (IsServer) StartCoroutine(DespawnCoroutine());
        }

        protected virtual IEnumerator DespawnCoroutine()
        {
            yield return new WaitForSeconds(deathDuration);
            Debug.Log("Despawned!");
            NetworkObject.Despawn(true);
        }

I'm using Multiplayer play mode and the enemy is destroyed on the hosts editor, but not on the clients
Is iti because the enemy is not spawned but is placed in the scene?

[sharp axle]

Scene objects should not be destroyed. If it need to be dynamically despawned then it should also be dynamically spawned.

[potent monolith]

Got it, so itd be betterto just make an empty that immediatley spawns my object if I want a despawnable scene object?

[sharp axle]

If it needs a spawn point, yea

[opal dust]

When using netcode for entities what's the recommended way to handle a FPS player camera? There doesn't seem to be a lot of examples or documentation out there for this kind of stuff so I've tried a few different things. I've tried having just 1 camera in my sub scene and then attaching that to the network spawned local player entity but the camera isn't positioning itself correctly (in fact inside the entity hierarchy it shows that I have 2 cameras after making it a child), I've also tried making the camera part of the player prefab and then creating a system that enables the camera for the local player but nothing is rendering even though I've added the "main camera" tag to the spawned camera. Would appreciate some help.

Sorry if this isn't the right place to ask because this is a DOTS / networking related question

[sharp axle]

You should check out the DOTS Character Controller sample. Its got 1st and 3rd person controllers. it takes a little bit of setup but works really well

[opal dust]

Thankyou, appreciate it

[weak plinth]

Anybody here familiar with Photon Server?

[vital hawk]

fholm works for Exit Games but I don't think he's involved in Photon Server specifically

[graceful zephyr]

I know photon server decently, worked with a fair amount

[safe coral]

Has anyone started looking into the new Unity Multiplayer alpha? https://github.com/Unity-Technologies/multiplayer

[amber adder]

I think the guy from BenchmarkNet github project summed it up pretty well. https://forum.unity.com/threads/jobs-any-way-to-access-static-data-safely.541070/#post-3987865

I have had the most success with the low level libraries such as Forge or LiteNetLib. Photon has a lot of hand holding which can get in the way of clean code and performance.

[safe coral]

Wow, quite the scathing review. Glad to see a Unity dev respond at least.

My angle is that I had been using Unity's low level API in prior and currently live projects. My hope was for a continued/expanded support in that area after the prior has been deprecated. I've never been fond of relying on third party support.

I was also hoping for a relationship with Google Cloud that benefited the developers but I can only see it being somewhat of an exclusivity deal from my limited perspective. Is there more information somewhere and what they are specifically bringing to the partnership?

[small fulcrum]

See https://unity3d.com/partners/google/connectedgames The most visible work done available through the collaboration at the moment is Open Match (https://github.com/GoogleCloudPlatform/open-match).

[graceful zephyr]

@safe coral i mean the guy is not wrong, the current 'alpha' of the multiplayer stuff on github is something anyone half decent at network programming could build in a couple of days.

And there's been no updates, no information, etc.

Other than the constant re-iteration of that they want a jobified network system, should work with the ecs, etc.

things we've all heard 50x times now

[warm geyser]

hi @graceful zephyr is that you the guy behind photon true sync ? what a pity such a great project suspended ! do you have any plan to put it back on track yourself ?

[graceful zephyr]

@warm geyser no i'm not, i'm the guy behind photon quantum

[warm geyser]

ah sorry, I made it wrong, good project anyway, Quantum is also very promising

[graceful zephyr]

or well we are a team working on photon quantum, but it's my invention/original code it's based on

[warm geyser]

but the price to access it is not cheap at all

[graceful zephyr]

I'm not solely developing quantum, we're a few guys working on it

I'm also the guy behind Photon Bolt, maybe that's where you confused it

[warm geyser]

yeah, Photon Bolt, got it

so limited choice today to find a determinism physics networking library in Unity to work with, I just can dig into the last version of true sync, not sure if I have enough brave to apply it for live game, it's deprecated

[graceful zephyr]

@warm geyser yeah we don't have support for true sync in our cloud anymore afaik

so you need to reimplement all of that also, i.e. the networking

@warm geyser again quantum is also deterministic, but takes a much different approach, and thus performs much much better (orders of magnitudes faster)

[warm geyser]

exactly what I'm also looking for, re-implement the networking part to use any 3rd networking solutions I want

quantumn is so expensive, can't afford

[graceful zephyr]

also i have no idea what the license for the true sync code is, as I never dealt with that

[warm geyser]

oh it can be used with PUN plus license normally

[lusty crow]

Oh hello fholm, nice to see you here!
Since there is now the quantum framework, which seems great,
and truesync is discontinued,
I hope it is still okay to use the deterministic physics parts of it, namely the box2d conversion and the q31.32 fixed point type?

I tried to make calculations much faster by ignoring overflows and underflows, but I guess this could lead to non deterministic behaviour? that kinda concerns me,
but the old implementation of truesyncs number system was sadly so slow that it is hardly usable for big games.

[graceful zephyr]

@lusty crow I can't answer if it's okey to pull pieces of the true sync code for other things

You need to ask via our support email so someone with knowledge about the licensing and such can answer

Just a coder, so don't deal with that stuff

[lusty crow]

oh okay, thank you!

[graceful zephyr]

Quantum uses 48.16 as it allows us to do the fast mul and div

Gives up decimal accuracy for performance

(a lot of performance)

As overflowing 48.16 is waaaay harder than 31.32

[lusty crow]

yes, there is one operation, I am not sure if it was multiply or divide, which would overflow 31.32 immediately,
right now I changed a few things in the base number class to have q40.24
this still allows me to have a good lower end precision and my maps are rather small.

and I got rid of the - using a BigInt as intermediate register to handle calculations and overlow

TrueSync had like 60 lines of code for performing a simple multiplication between 2 fixed point numbers

[glass bobcat]

Hi! I've been learning networking for a while now and I was wondering if any of you had some good networking books relating to games?

[ornate geode]

Seen this one yet @glass bobcat ? https://github.com/MFatihMAR/Awesome-Game-Networking

[glass bobcat]

Yup I have it bookmarked, p sure you sent it to me on the C# server last time :^)

[ornate geode]

Didn't i send you the Unreal Engine one?

[glass bobcat]

I have a book for that one

i like the framework

it does just about everything for you lol

[ornate geode]

Didn't i send you this one: http://cedric-neukirchen.net/Downloads/Compendium/UE4_Network_Compendium_by_Cedric_eXi_Neukirchen.pdf

[glass bobcat]

that's the one I have printed

[ornate geode]

https://www.oreilly.com/library/view/multiplayer-game-programming/9780134034355/ This one was also pretty nice

[glass bobcat]

i'm building my own lib, I just need pointers on what's "standard" for devs, like the normal patterns and stuff

oooh nice

[graceful zephyr]

@glass bobcat low lvl UDP or higher level stuff?

[glass bobcat]

oh hey @graceful zephyr, it's been a while since we talked ^^ I like both as long as it's UDP

[graceful zephyr]

Hey

No I mean what do you need pointers on how to solve

[glass bobcat]

General programming patterns for networking libs

Packet serialization, I understand it but I'd like to see a good version to compare to what I wanted to do

Ticks too

[junior finch]

@glass bobcat This guy has a bunch of great articles on the subject, https://gafferongames.com/

[glass bobcat]

I've been reading his articles

They're nice

@graceful zephyr i'm trying to create a networking lib so I guess general low level UDP stuff

[graceful zephyr]

look at glen fiedlers repo on github

netcode.io on github

that has a good socket impl

[glass bobcat]

👌

Is it bad that I'm trying to abstract a ton of things @graceful zephyr ? I'm basically letting the lib's user either use my implementations or switch them over for his

Should I just have a concrete architecture built in, and let the lib's user use methods and hook to events?

[vital hawk]

you mean this one? https://github.com/networkprotocol/yojimbo @graceful zephyr

[graceful zephyr]

@vital hawk https://github.com/networkprotocol/netcode.io

[vital hawk]

oh I thought you meant two different things 😄

[glass bobcat]

var builder = new ComponentBuilder()
    .WithSocket<UDPSocket>()
    .WithPacketHandler<NetHandler>();

Is this a bad idea? I'm basically letting the dev choose what class to use for each "module" of the lib, and then Build() the lib accordingly.

[graceful zephyr]

@glass bobcat just seems a bit over engineered

Just write out a lib that works then first time

I'm not much for abstraction for abstractions sake

[glass bobcat]

Alright

[celest hamlet]

Hi! Anyone knows if using an WWW request in a web with https automatically encrypts the data sended/received?

[last hornet]

does the new TOS stuff that hurt SpatialOS basically ban any form of unity based dedicated server?

https://forum.unity.com/threads/recent-tos-update-blocks-the-use-of-spatialos-to-make-games-in-unity.610447/

all the TOS language is worryingly vague

[errant tulip]

"You may not use a third party to directly or indirectly distribute or make available, stream, broadcast (through simulation or otherwise) any portion of the Unity Software unless that third party is authorized by Unity to provide such services."

This alone is a bit ... much.

[last hornet]

yeah if you take it at face value it seems to imply you couldn't even sell your game on steam

[errant tulip]

So I guess it also makes sense to decouple your networking / backend as much as possible from Unity.

[small fulcrum]

No it does not basically ban any form of Unity based dedicated server. There will be some public clarifying language later today

[errant tulip]

Thanks for the update.

[graceful zephyr]

@small fulcrum thanks

[jagged radish]

Will it be posted on social media like Twitter?

[small fulcrum]

I honestly don't know what channels it will be posted through. I expect Twitter. I'll post it here too.

[vital hawk]

I'll just continue to work on my now multiplayer again game then 🤔

[glass bobcat]

neato

[weak plinth]

Guess I'll as well

[olive coral]

When we have an update to share with you regarding SpatialOS we will make sure everyone in the Discord knows where to find it.

[jagged radish]

right now people are panicking so hope this will get clarified soon

[pure atlas]

Yeah I have a bunch of dev friends right now working on multiplayer games who are trying to decide whether to just give up their games or transfer over to a different platform

[jagged radish]

Our projects are multiplayer games so... we are on the tightrope right now

we were working for years right now on them

[pure atlas]

yeah luckily my current project is singleplayer

but i feel for all you guys just having the rug swept out from under them

I'm also curious to know how unity will treat stuff like Steam Links

since that's technically a streaming service as well (albeit locally)

[jagged radish]

I don't think that will be affected because its local

[pure atlas]

the problem is that the stuff says "over the internet or other network"

"network" is a pretty broad term

[jagged radish]

ugh

[pure atlas]

right?

[graceful topaz]

I wouldn't be too worried, I doubt Unity would want to hurt developers on their platforms to showcase how good their platform works, and especially the games that are made from it.

[pure atlas]

the problem is that they already did the damage

they already did hurt devs

[glass bobcat]

Just wait for them to clear things up today

Might just be badly worded

[pure atlas]

unity is a pretty good engine, but there are other great alternatives and I know a lot of devs will be asking themselves after this "why work with a company who clearly doesn't care enough about the community to even make sure they get their legal jargon correct?"

it's not like they're some tiny little indie dev company who doesn't employ any legal team

and im sure the legal team they have is incredibly good.

this whole situation just leaves an awful taste in my mouth, and i don't want to have to worry about making a game and unity pulling the rug out from under me because of an abrupt ToS change that we weren't expecting

[meager dragon]

Unity didn't make the post. So how can they be held liable for it's verbage?

[pure atlas]

im referring to the ToS change's verbage, sorry.

[bright saffron]

They wrote the ToS

[meager dragon]

I was referring to the blog post and Reddit post that made everyone aware

ToS last updated December 5th

[half tree]

https://twitter.com/SpiltMilkStudio/status/1083419415247572992

Spilt Milk 💦 (@SpiltMilkStudio)

UPDATE: Lazarus is back up.

We're not quite sure what's going on - we were told that access to the servers would be revoked by 2:30pm today but it seems that's not the case.

Until either the servers are forced down or we’re told to turn them off, we’ll keep Lazarus l...

Twitter

[weak plinth]

what's the difference between LiteNetLib and libgren

[vagrant gazelle]

I'm not following the networking threads right now but does Unity just banned every third party multiplayer solution out of nowhere!?

[split crescent]

They're working on an update to provide some clarification

[graceful zephyr]

@vagrant gazelle no they didnt

[vital hawk]

@vagrant gazelle that's what people panic it's about

[graceful zephyr]

@weak plinth LiteNetLib is good, Lidgren isn't.

😄

[vital hawk]

they added extra wording to their ToS that caused some heat as people don't really know how Unity meant it to be interpreted

but it'll be cleared out soon

[pure atlas]

people don't really know how Unity meant it to be interpreted

if that ain't the legal system in a nutshell

[weak plinth]

whats the difference@graceful zephyr

[graceful zephyr]

i said

[weak plinth]

what i meant is why

why is litenet good and lidgren bad

[pure atlas]

because he said so. duh!

(in all seriousness, i'd like to know as well)

[graceful zephyr]

because lidgren is old and outdated, is horribly and overly complex internally, has a very wastefull ack/nack system etc.

[split crescent]

think folks misinterpreted your comments as being TOS related

[graceful zephyr]

oh TOS?

[split crescent]

or atleast I did

[graceful zephyr]

nah

was talking about lidgren/litenetlib

[split crescent]

I googled it to try and figure out how it could possibly be related

potentially need more coffee over here

[graceful zephyr]

lol

im a bit scatterbrained today so maybe my sentances dont make much sense

[torn sage]

https://i.redd.it/pk8s45v8yn921.jpg

[split crescent]

Blog linked in #497866432277643275

[torn sage]

👍

[jagged radish]

👌

[graceful zephyr]

The blog post 100% contradicts the ToS, problem being the ToS is a binding legal agreement and a blog post is not

[graceful zephyr]

But epic and improbable now starting some game dev fund of 25 mil USD together seems fishy as fuck, they managed to come together on that deal together in 12h? This seems awfully coordinated on their end

[graceful zephyr]

Honestly from what I've seen i got to say I'm on unity's side here, yes the ToS/EULA need to be cleared up, but it seems like improbable had a year to solve this issue and didn't, and know they cozied up with epic to fling shit at Unity

Not that anybody cares about what I think 😄

[vagrant gazelle]

I have no idea what's going on...

[glass bobcat]

i've seen the term Networker a lot in networking libraries, what exactly is a networker?

[brazen basin]

Like a librerian

[glass bobcat]

[gritty sail]

Librarian*

I think he means to say that it’s the part of the library that handles the networking like how a librarian handles the books in a library

A fairly loose allegory as any profession could do

[glass bobcat]

i get it, but I will need more details :^)

[graceful zephyr]

@glass bobcat there's no well defined meaning of that term

[glass bobcat]

o

okay

[mild rampart]

@graceful zephyr About that TOS stuff, I totally agree with you, that dev grant thing is super fishy. I guess they gotta use that fortnite money for something lol

[weak plinth]

I have a idea

Why not use it on the documents

That's what literally everyone wants

[restive urchin]

The did use it on back paying asset developers last year. THat was v cool for a few of my mates 😄

I think they like throwing money around 🤷

They were also giving out a lot of Vives a couple years ago

[valid coyote]

They give out several sizeable grants a year to small developers too

They have a lot of money and aren't afraid to use it to try and help developers

[glass bobcat]

Hey, quick question: does the client simulate at the same pace of the server, or can it simulate faster?

And how would that work, do I send deltaTime to the server per simulation?

[gritty sail]

Normally a client requests data from the server, so usually the server runs faster

[glass bobcat]

It's more about how many times per second the client should send input to the server

i feel like the client would pool as fast as it wants

as long as it's at least as much as the server it shouldn't matter

[gritty sail]

Depends on the input i would say, if its something like doing gui that has backend then send it on event, but i guess you are talking about continuous input such as player movement, i would recomend sending that as fast as the server can handle it

[graceful zephyr]

They simulate at the same pace usually

Usually a fixed tick rate like 60

And they both stick to it

[glass bobcat]

what about low tick rates

does the client really simulate only 20 times per second?

i mean, it might not be that noticeable actually 🤔

[graceful zephyr]

Low tick rates?

Don't mix up simulation tick rate and packet send rates

[glass bobcat]

oh right

should the client request the simulation rate from the server?

[graceful zephyr]

Nah that's usually decided in code and fixed for both

But sure it can

If you want variable tick rate

[glass bobcat]

👌

what about packet send rates, same as tick rate or should it vary?

[graceful zephyr]

Depends

[glass bobcat]

packet send rate can be lower than simulation rate right?

but not vice versa?

[graceful zephyr]

@glass bobcat generally yes correct

[glass bobcat]

thanks

[leaden grove]

Hello! I'm using a fork of UNet and I'm having a bit of troubles telling a player that has joined late the state of the game. Sending a TargetRpc to the newly joined client doesn't seem to work. I haven't tested having a SyncVar hook because I suspect that clients already in the game could be dragged into the hook because their "started state" wouldn't be set before the hook fires.
Anyone that has any smart ideas how you could go about telling them if the game has started, ended, etc?

EDIT: Apparently it was just as easy as checking the state in Start on the client that just joined.

[loud vapor]

Hey guys. So with Unet being deprecated and if I wanted to play around with networking in unity right this very second - what terms should I google to help me get started?

or perhaps a section of the docs I can refer to?

even in the 2019.1a version of the docs I believe it's still just talking about the deprecated unet apis

[compact bramble]

I recommend Photon as a stable service that isn't infringing anything, until Unity's new networking comes out of Alpha.

I don't recommend anything else because it's all a wild west.

[loud vapor]

Okay so it's third party or wait for unity to catch up because we no longer believe in p2p

[compact bramble]

p2p meaning, specifically relay server?

or actual person to person (cos almost nobody does that)

[loud vapor]

I mean, unity wants us to move away from p2p and into dedicated hosting only?

[compact bramble]

No, they don't.

With all of Unity's networking so far, you have never been forced to choose a specific thing.

I used to run Unet Peer to Peer just fine.

Unity will supply cloud relay (this is where one player is client but sends messages to a server which relays back and forth to other clients) - the reason for this relay (that most games do) is to reliably deal with firewall problems, dropped hosts (migration) and a consistent service.

Is that what you want, or do you want dedicated?

Unity will also do the whole dedicated thing but its a way off yet

[loud vapor]

I have no clue what I want I just wanted to see what the process looks like

[compact bramble]

Ah well just get PUN (photon unity networking) - a free one is fine if you can - and this will basically be just a "relay" server, which means one player is also the server, just nobody directly connects to each other (which is a pain without ip addresses and firewalls and nobody deals with that in modern games)

[loud vapor]

peaches.

[compact bramble]

I wouldn't touch UNet though, it's simply not good enough IMHO

[loud vapor]

thanks for the info hippo

[compact bramble]

np

Can probably migrate later since PUN is kind of a basic and well known approach

[graceful zephyr]

@compact bramble hey fun to see you here o/

[compact bramble]

hey you too 😃

[graceful zephyr]

@compact bramble whats up

been seeing you around on the ecs forums

i dont post super much myself there tho, but read most things

[compact bramble]

Yo, I'm just developing as usual, same old. how about you?

[graceful zephyr]

@compact bramble ditto, networking code as always

working on photon quantum, fiddling with my hobby libs, etc.

the usual 😛

[compact bramble]

Photon Quantum sounds cool. I generally promote photon at the moment, I don't use it but it's by far the best general purpose option I think

I will use Photon if Unity's own stuff is not ready by the time I need to implement networking...

[graceful zephyr]

@compact bramble quantum has been/is great, we dont promotive on asset store or to hobbyists really, just focus on studios/aaa

[compact bramble]

Yeah understandable since you get paid then 😛

[graceful zephyr]

its my invention, and we turned into a product together with the photon guys

[compact bramble]

oh tell me about it?

Should I be switching to Quantum?

(wasn't it Bolt?)

[graceful zephyr]

depends, so quantum is a deterministic gameplay framework

its not just a networking system

it's a completely custom gameplay loop/framework, all deterministic, made to run inside (and outside of) Unity

has support for full predict/rollback of the game on the client

so you cant really 'add' quantum to your game

it has to be built IN quantum

here is a video of the current game im building in it to try to push the boundries

so it has a custom physics engine, custom ecs, custom nav mesh, etc. custom everything

all hand made to be deterministic

[split crescent]

custom physics since physx isn't deterministic?

thats getting addressed at some point, right? 😛

[graceful zephyr]

@split crescent not just that, physx doesnt support predict/rollback either

[split crescent]

got it

[graceful zephyr]

so both need to make it deterministic AND be able to predict/rollback the entire game state

[compact bramble]

It looks good but does it qualify as a platform? Photon should be talking to Unity in light of the recent SpatialOS fiasco

[split crescent]

presumably they're not selling servers running it

and just the lib

[graceful zephyr]

@compact bramble we only host the relay servers

[compact bramble]

OK then np there

[graceful zephyr]

like we always do

[split crescent]

very interesting on quantum - thanks for sharing

[graceful zephyr]

you can host the game inside of Photon if you choose to self host photon server

so you can get an authoritative result of the game

since quantum can run outside of Unity also it can host a quantum instance inside of a photon room

[compact bramble]

This seems like an absolutely enormous undertaking creating a networked deterministic physics-based engine, basically

[graceful zephyr]

yes, it was :p

[split crescent]

what's your target deploy? .net core?

[graceful zephyr]

@split crescent anything that photon runs on, officially atm i thin we only support windows/.net framework

but i know theres work being done on other runtime platforms

[split crescent]

gross

😄

[graceful zephyr]

@split crescent not as gross as mono 😄

[compact bramble]

So it's definitely a separate product, and not an evolution of PUN?

[graceful zephyr]

@compact bramble correct, it is not

it runs ontop of our low level API called Realtime

you can use it WITH PUN if you want tho

they are compatible

[compact bramble]

What does the code look like to get something moving (from the customer developer POV, not yours)

[graceful zephyr]

but its not going to replace pn

@compact bramble how do you mean?

like creating a sphere and pushing it around?

[compact bramble]

yeah!

[graceful zephyr]

2 sec, let me bring up github

[compact bramble]

I'm just vaguely curious how it looks

[graceful zephyr]

    const int h = 2;
    const int w = 2;

    public void DropSpheres(Frame f) {
      for (int x = -w / 2; x < w / 2;) {
        for (int y = 6; y < h + 6; y++) {
          var box = f.CreateBox();
          box->DynamicBody.InitDynamic(Core.DynamicShape.CreateBox(new FPVector2(FP._0_50, FP._0_50)), 1);
          box->Transform2D.Position = new FPVector2(x + f.RNG->Next(), y + f.RNG->Next() * 2);
        }
        x += 2;
      }
    }

@compact bramble from some random physics test code i had laying around

on laptop atm

[split crescent]

FPVector => fixed point?

[graceful zephyr]

ye

[split crescent]

hand rolled or using something like fixed.net?

[graceful zephyr]

hand rolled to the t

[split crescent]

cool

[graceful zephyr]

full deterministic math lib, all math functions, massively optimized

[split crescent]

we're using Fixed.net for project I'm on - kinda wish I just used decimal tbh

[graceful zephyr]

we take about a 30% hit on top of float performance, so its pretty ok

some stuff is faster than float, but on avarage its a bit slower

[split crescent]

yeah

Fixed.net a real GC pig

[graceful zephyr]

we have zero runtime allocations usually

once the game has spun up

and created everything

its all built with native memory, pointers, linear laid out mem blocks, etc. ( @compact bramble sounds like unity ecs? :p)

[compact bramble]

iq has a nice blogpost on rationals if you want to read through it : https://iquilezles.org/www/articles/floatingbar/floatingbar.htm

[split crescent]

whats a "box" in above snippet?

Analog a GO?

[compact bramble]

(and yes it sounds a bit like Unity's ECS but nothing stops you..) 😃

[graceful zephyr]

@split crescent no thats just a box collider

[split crescent]

got it

[graceful zephyr]

big difference vs. unity ecs is that

1. we expose the pointers to the user, as we dont target this for the avarage unity user but experience teams this works out well (less copying, etc.)
2. our job/threading system is not as good :p

also since agian, its not a generic for any type of game framework

[split crescent]

How long have you guys been working on quantum?

[graceful zephyr]

we've made mode tradeoffs, and we have some more limitations imposed due to needing to support full predict/rollback

@split crescent I developed the original code base around 2014-2015, i've been working on it with exit games since early 2016

[split crescent]

cool

[graceful zephyr]

a lot of weird problems to solve when you're trying to a lot of players in a deterministic predict/rollback environment

[split crescent]

yup

Have you (or anyone else) been in touch with Unity as they've been fleshing out V3 of their networking libs?

[graceful zephyr]

I've talked to some of the unity networking guys

but only on a personal level

[split crescent]

feels very hypothetical atm

not sure how far along they are

[compact bramble]

The alpha is available

[graceful zephyr]

    static FPVector3 FindRailgunTarget(Frame f, Ship* ship) {
      using (var hits = f.Scene3D.Raycast(ship->Transform3D.Position, ship->Transform3D.Rotation * FPVector3.Forward, ship->ShipType.RaygunRange)) {

        // sort based on distance
        hits.Sort(ship->Transform3D.Position);

        // find closest hit that is not ourselves
        for (Int32 i = 0; i < hits.Count; ++i) {
          if (hits[i].Entity != null) {
            var entity = (Entity*)hits[i].Entity;
            if (entity->EntityRef == ship->EntityRef) {
              continue;
            } else {

              if (entity->Type == EntityTypes.Ship) {
                var entityShip = Entity.CastToShip(entity);

                // damage ship
                Ship.Damage(f, entityShip, ship->ShipType.RaygunDamage, hits[i].Point);

                // push ship
                entityShip->DynamicBody3D.AddLinearImpulse((entityShip->Transform3D.Position - ship->Transform3D.Position).Normalized * ship->ShipType.RaygunForce);
              }

              return hits[i].Point;
            }
          } else {
            return hits[i].Point;
          }
        }

        // last resort, hit nothing
        return ship->Transform3D.Position + ((ship->Transform3D.Rotation * FPVector3.Forward) * 1024);
      }
    }

here is some actual game code

from that space game

for firing the railgun

[split crescent]

1024

[graceful zephyr]

@compact bramble @split crescent i mean the current alpha is... just a socket lib

@split crescent haha, yeah noticed that now... need to pull the correct value from the ship config

and im kind of sad to say its not even a good socket lib, its just the bare bones that anyone with half decent networking experience can build in a week

not even sure why they released it

like, i wish it was better, i could build assets for it on the store that would be awesome, but in its current state

it's... i mean, yeah... it boils down to this:

1. those of us that know enough to use it, can make something better ourselves and not run the risk of having an alpha change API or being pulled
2. those that cant use it.... well yeah.

[split crescent]

so in above, do you have traditional GameObjects w/ transforms in the view that are observing the underlying model representations?

[graceful zephyr]

dont judge the code for that spacegame too much, it was built in 3 days to try to push the limits

[split crescent]

or are you taking a diff approach to draw?

[graceful zephyr]

@split crescent so the simulation is completely standalone from Unity, COMPLETELY, it doesnt even hold a reference to UnityEngine.dll (this is what lets us run the game on the servers, etc.)

(We of course have tooling on the unity side to export stuff into the simulation)

Unity has full access to the simulation state as the game is running, how you choose to render the simulation state is up to you

Unity basically becomes a rendering front-end for the quantum simulation

[split crescent]

yup

got it

[compact bramble]

It would be creating an ECS world to handle the data from your own ECS implementation

[graceful zephyr]

We obviously have some pre built tooling to make it easy to get started, etc.

Like we have a custom 'hookup prefab component'

and stuff like that

[compact bramble]

ah cool

[graceful zephyr]

but generally the serious games that are built with it

do their own thing

that fits their game perfectly

@compact bramble I've played a decent amount with the Unity ECS, it's getting there but has a long way to go in my opinion

I've made networking lib that puts all the game state in the unity ecs, but still lets you build the game using GOs, etc.

So dug pretty deep into the ECS, and really used it for something real

It's fast, absolutely (especially on my 32 thread cpu :p)

[compact bramble]

It will take Unity quite a while to make ECS versions of all their current functionality

[graceful zephyr]

I can serialize and dispatch 100k entities out to the network in like 1 ms

which is pretty fucking good lol

[compact bramble]

yeah lol

In most mp games you're spending probably most of the time on setup

[split karma]

not for your internet connection 😄

[graceful zephyr]

@split karma 😛

@compact bramble but a lot of problems in the ecs currently, are very hard to solve

because some thing are super easy by just doing 'hold a reference to this thing'

and in the ecs you end up jumping through infinite hoops to get it done

[split karma]

its usually also a decent amount of hoops to go through for the stuff that you can solve

[graceful zephyr]

I understand WHY its like that, hell we built our own native memory backed ECS for quantum

but that doesn't mean its a good experience for the users of the ecs as it is today

[split karma]

ah ok so you have ECS for quantum, but the user pretty much never has to actually touch it unless they want to right?

thats basically what Unity is missing tbh

[graceful zephyr]

@split karma no the ecs is baked into quantum, you dont HAVE to use it, but you pretty much should unless you want to build your own determinsitic setup from scratch

[split karma]

ah

[graceful zephyr]

but quantums ecs is not a 'generic gameengine ecs' for any type of game

it's specifically targeted at higher end studios, with good programmers, etc.

[split karma]

yeah that makes sense

[graceful zephyr]

that know how to build stuff by hand if need be, etc.

and again, i am not complaining over unitys ecs, i think it's great, but it is aiming to be a generic game engine solution for programming all games basically

for the avarage unity users

which is a much higher bar

[split karma]

i dont know, they may put abstraction layers on top for the average user too

[graceful zephyr]

yeah sure, but that's what i mean

it's not there yet

[split karma]

yeah

[graceful zephyr]

they will do that in the end, but the bar is much higher

i can expose low level pointers and allow people to fuck with memory by hand

[compact bramble]

"it's specifically targeted at higher end studios, with good programmers, etc." because anyone smart enough to make Quantum is smart enough to avoid even contemplating the depths, the horrors of supporting customers that aren't... or basically 99% of the asset store.

[graceful zephyr]

because the people using quantum know how to do that, and they want that ability/power

@compact bramble yeah i did that once with Bolt, never again.

never. ever.

I still get support emails for BOlt, 3 years after i sold it to exit games

[compact bramble]

I never published to begin with. I knew. I smelled it.

[graceful zephyr]

got one just this morning

[split karma]

I mean who better to contact 😛

[graceful zephyr]

no wait... its almost 4 years since they bought it lol

may 2015

so yeah

never. again.

[compact bramble]

Well thing is at least with Quantum they can get both an answer and no answer at all.

[graceful zephyr]

We have about 50 studios atm working on games in quantum

some larger studios like ubisoft

[split karma]

wow thats more than I expected

[graceful zephyr]

some smaller indie shops

[compact bramble]

That's a lot, way more than expected. I'm surprised because - it's a lot of trust for something unfinished.

[graceful zephyr]

quantum isnt unfinished :p

[compact bramble]

ECS is not a finished problem 😛

[graceful zephyr]

true

i think thers like 10-15 games launching during 2019 at least

built on quantum that is

dont quote me on that i dont keep up to date with all the customers schedules, etc.

[compact bramble]

Full SDK-Access per month:
US$ 1,000

Including in-depth instructions by our authorized Quantum engineers

(or just pop in discord and flirt with fholm but don't tell anyone)

[graceful zephyr]

@compact bramble ye, lol :p we have our own locked down discorder server for day to day support

its been working out pretty ok

its fairly low volume

[compact bramble]

"We are searching for great teams who want to leverage the full potential of Photon Quantum.

Are you one of them?"

I could be but really, this solution actually looks like it needs someone on networking full-time and not a sole-programmer like me who is split between a whole bunch of different jobs

[graceful zephyr]

well so the thing is right

because its deterministic

there's no networking code to write 😛

thats the whole selling point

[compact bramble]

This basically seems like the same thing Unity wants to offer, from what Unity's said publicly so far, under the "by default" thing

[graceful zephyr]

well they are still talking about classic networking tho arent they

just that if you write it in a certain way

it will network 'easier'

you still have the same regular concerns, etc.

i doubt they will make the whole engine deterministic with predict/rollback

but idk any about their inner workings ofc

[split karma]

I think determinism for ECS shouldn't be an issue

[graceful zephyr]

@split karma oh sure, probably not

but predict/rollback determinism for the whole game state is a much much harder problem

not saying they wont or cant do it, absolutely, they have way more resources

[split karma]

but yeah prediction and rollback are a different kind of challenge to shove into something like it

[compact bramble]

and potentially quite expensive for a traditional approach

[graceful zephyr]

but it's a weird take to make all games fit in

i'll be the first to admit that quantum isnt suitable for all games

determinism costs performance, a lot of it, even if they make floats deterministic

[compact bramble]

Well the concept that it basically needs no networking code puts it on the radar for practically any realtime mp game.

[graceful zephyr]

@compact bramble yeah, you write the game as if it was a singleplayer or split screen game

and it works in networked online play, singleplayer and split screen automatically

the game itself doesnt even know its networked in fact so

[compact bramble]

No 3D physics is a bit of a blow

[graceful zephyr]

we do have a 3d physics engine now, but it's in alpha/rudimentary state

thats what im using in the space ship video

it does work, and will do fine for something like the space ship game

[compact bramble]

This is something people should be looking at when they begin the game's development

[graceful zephyr]

but if you want to do some full stuff with joints and wheel physics and crap, no it cant do that

@compact bramble yeah, we generally only sell to teams starting out development

as it's a massive undertaking to port the game

some teams have, because they tried some of the demos we have that were similar to their games and played them online

and its basically flawless, and you can do so many things which are super difficult when dealing with non deterministic systems so easily in quantum

[compact bramble]

This is a year old but still exciting: https://www.youtube.com/watch?v=b9T-k-R7rKs

[graceful zephyr]

so some teams did port their games to quantum, but i'd say 9/10 teams we got in earely

@compact bramble that game is one of the first games built on my original code base, made by my brother and his friend in two weeks.

[compact bramble]

sweden to brazil is impressive in what looks like no latency

[graceful zephyr]

https://steamcdn-a.akamaihd.net/steam/apps/256666769/movie480.webm?t=1468310068 heres the first game they built, in about 2 months

[compact bramble]

Usually one does not even bother talking about gaming that far apart online

[graceful zephyr]

@compact bramble yeah it offers insanely good performance even over high ping scenarios

@compact bramble this is the space game with both players having a ping of 500 ish

the client im not focusing is a bit jittery, thats just the rendering getting shit on in the non-focused video when the recorder is running

heres an even older with around 230 ping or so on both clients

you'll see when the ship blows up after the first few seconds how it basically happens at the same time on both screens 😛 despise there being like 300ms of total delay between them

@compact bramble but yeah so thats what i've been up to, what game are you working on?

[compact bramble]

that looks really impressive though

ignoring any issues

(Offtopic - my game is coming along well, but slow because it's self-funded... rpg lite action game with a fair amount of comedy to it, physical toys etc)

[graceful zephyr]

@compact bramble ah nice, like third person rpg or top down sort of thing?

[compact bramble]

3rd

and first

[haughty vigil]

Hellooo, can anyone either tell me how, or maybe point me to (with a link) where i find out how to; setup multiplayer in Unity, without using the Network Manager, that apparently is deprecated and going out of "business" 😅

[midnight tree]

We used Photon, and later on, Quantum SDK for our PvP title

You can try it out @haughty vigil

[flint spade]

If I were making a LAN based mobile controller thing (you would be able to use your mobile as a controller) which protocol would be better TCP or UDP?

[gritty sail]

TCP

[last hornet]

@midnight tree i don't think you can just "try out" Quantum 😃

[compact bramble]

yeah it's a noob engine anyway (just kidding @graceful zephyr ) 😃

[graceful zephyr]

@compact bramble ;p

I think he meant the game

[glass bobcat]

if it's LAN only TCP, and I don't see why you couldn't use UDP in that case

UDP would probably work and have less overhead

if you plan to make it anything else than LAN go with UDP

[last hornet]

i mean it is technically a noob engine

thats why its so attractive

just make yo game

[weak plinth]

@haughty vigil https://github.com/Unity-Technologies/multiplayer/blob/master/com.unity.transport/Documentation/TableOfContents.md

[weak plinth]

i was wondering if this is basically how you'd go about separating client / server code and running both for testing:

public class Game : MonoBehaviour
{
    List<IGameLoop> gameLoops = new List<IGameLoop>();

    void Awake()
    {
#if UNITY_EDITOR
        gameLoops.Add(new ServerGameLoop());
        gameLoops.Add(new ClientGameLoop());
#elif UNITY_SERVER
        gameLoops.Add(new ServerGameLoop());
#else
        gameLoops.Add(new ClientGameLoop());
#endif
    }

[vagrant gazelle]

Wouldn't it be easier to just do

#if !UNITY_SERVER
// some code
#endif
#if UNITY_SERVER || UNITY_EDITOR
// some server code
#endif
```It will not require whole IGameLoop logic

[weak plinth]

i'm honestly not sure yet

would you be sprinkling those all throughout the codebase?

[vagrant gazelle]

Probably

[weak plinth]

mm, i'd have to think about that. i guess it depends on how much overlap there is?

[vagrant gazelle]

I think that for testing better way would be to create a tool that starts a server as soon as you enter play mode

[weak plinth]

yeah that's probably true

[vagrant gazelle]

I think that FPS Sample has something like that but in more advanced form

[weak plinth]

yeah i had taken some of the game loop ideas from it, but the project tools panel was pretty over my head so far

[split crescent]

I think in general, you should treat forking code like that

IF_SERVER
IF_CLIENT

As a code smell if its doing other than switching out the implementation of a manager or a service

[vagrant gazelle]

I haven't done lot of networking myself but I don't think that doing basic tool for testing would be hard. Something like

if gone to playmode
   run "project/server/server.exe -maybeSomeParameters"

[weak plinth]

yeah, i think i'll have to give it a shot

[vagrant gazelle]

Try Process.Start(...), that will start an executable (even with parameters if needed) as far as I know

[weak plinth]

will do, this'll be my first editor extension so gotta read up on the ol' docs

[vagrant gazelle]

Maybe using Process class itself would be better, that way you will be able to terminate the server when you'll go back to edit mode

[weak plinth]

i dont know if this is networking per se but how do i go about saving game data so its not lost if the user uninstalls or changes devices. does unity offer something like that or do i need to have a service/host my own server?

[vagrant gazelle]

I think you need your own server, you can use Steam Cloud or Google Apps

[weak plinth]

im following this guide here, im not sure how much of this i should be understanding as a week 1 unity user

https://www.youtube.com/watch?v=0ECRLY2A_-c

[weak plinth]

What's the replacement for unity matchmaking system?

Dedicated servers?

[split crescent]

Will Photon Quantum eventually support 3D?

[vital hawk]

@graceful zephyr ^

[graceful zephyr]

@split crescent we have a basic 3d physics engine as it is today

But it's nowhere near physx or anything like that

It's for basic 3d collisions as of now

[split crescent]

cool - thanks!

[weak plinth]

any experience with using Protocol Buffers or MsgPack in Unity? mostly wondering if it worked out well and if there were any issues building to mobile platforms

[split crescent]

We're using protocol buffers fairly extensively

for mobile

a few things to note (you should be using proto3 and not proto.net btw as the latter is deprecated)

1.) C# libs weren't built with games in mind and can generate a decent amount of GC when parsing. The serialization savings are there which is nice, but it generates a bit more garbage than I'd like
2.) Proto3 does not support deserializing into structs, which is unfortunate
3.) Unlike other languages proto classes in C# are mutable. This can be a pro or a con depending on your use case
4.) C# classes that are spit out are partials, which is fantastic for extending data classes with business logic or convenience methods
5.) Be sure to compress ontop of serialize - the savings are meaningful
6.) Enums can't be keys in maps, (dictionaries) which sucks
7.) Be extremely careful with logging any proto message. If you do IMessage.ToString() you'll trigger an expensive reflection cost that attempts to look at all your proto messages in the current assembly. For us this was an 8s operation on an IPhone 5 and generated 6 mb of garbage.

still very recommended

didn't seriously consider msgpack, but we probably should have

took a look at flatbuffers, recommend avoiding

[weak plinth]

thanks a ton, I'll keep all that in mind

[torn sage]

Do you know if the UNET match making still works (2018.3). I know it is deprecated but it easier for me to use it for my proto.
When I create a match. Few seconds later it is no more joinable 🤔

[graceful zephyr]

@split crescent we are building an tps with the quantum 3d physics, and thats pushing just up to the limit of what it can do

[flint spade]

Would it be a better idea to compress class info with ProtoBuf/ZeroFormatter before sending it over the socket rather than just JsonSerializing/BinaryFormatting it

[split crescent]

Protobuf will generally serialize/deserialize significantly faster than JSON

[graceful zephyr]

@flint spade yeah

json/binaryformatter suck

[split crescent]

@graceful zephyr presumably with no height?

[graceful zephyr]

@split crescent no height?

[split crescent]

sorry, potentially a dumb question - it's a TPS, but presumably not full 3d

i.e. doom, not quake :p

[graceful zephyr]

@split crescent no its full 3d

[split crescent]

got it

would you consider it for a battle royale TPS?

(so hot right now)

[graceful zephyr]

@split crescent don't know if it'll scale to 100 players

[split crescent]

got it

[graceful zephyr]

@split crescent it's a hero shooter (overwatch) style sort of thing, i'll see if i can get an OK to record a video and show here

[split crescent]

cool - appreciate the context. No sweat if you can't 😃

[weak plinth]

i've been experimenting with Unity's new multiplayer code + taking hints from the FPSDemo that was released, but i'd be interested to know if anyone uses other network libraries, like ENet, ForgeRemastered, etc. i don't really know what the "gold standard" is and i'd kinda like something that's pretty well tested and has a good community

[compact bramble]

Photon is the most widely-used I think?

Networking for me, is about managed reliability, so a service that will not break because I don't have a decade of experience in it

[weak plinth]

right, the more i work through doing this the more i'm realizing i'm not going to be able to make the best networking for my game and then actually get to work on the game in the near future. Photon is definitely something i'm considering

[graceful topaz]

@weak plinth Honestly the best thing to do would be to just test the libraries yourself to figure out which one suites you needs better. As for Forge Networking, I would only recommend it to those who are willing to write some of the networking code themselves to get into the nitty gritty details. It handles some basic functionality out of Unity and holds your hand in that process, but if you want to do complicated networking it would only be a good networking solution if you planned on writing your own one day. Otherwise I would recommend using others out there.

[weak plinth]

gotcha, thanks for the input. i will have to start testing all of these out

[compact bramble]

It's not just a matter of getting computers talking it's also - can you maintain your own relay server? can you do this in the areas your game will be sold? What if the server gets caught in a DDOS?

will you be willing to pay those fees?

lots of considerations but if it's peer to peer or managed in steam you can get away with your own

[graceful topaz]

Indie games shouldn't worry about those problems as much and should just be focused solely on their game and making it fun. Bigger companies have bigger problems and smaller companies have smaller problems. Although if those are concerns for some people then they should indeed research them. I would play a fun game as long as it had singleplayer as an alternative. If the sole focus is multiplayer then it can indeed be a bigger concern as that is what everyone would mostly play.

[compact bramble]

when i stopped worrying and accepting that, started finishing 😄

[weak plinth]

thanks guys 😄

[stray knoll]

Hye there, I'm currently in the planning stage for our predominantly online game, and I'm revolving around a few concerns...

I'm gonna start by giving some more context about the game: It is a 2-player dungeon crawler, meant to be played either locally (split screen), over LAN, or over the internet. There's no singleplayer mode, so networking can't come as an afterthought. Multiplayer will be session-based and we expect games to last no more than 1 hour. Our primary target platform is PC but we'd like to try and get to Xbox and PlayStation. We're not sure yet if we will have enough funding to support a dedicated server architecture, but at the same time, we don't want to corner ourselves in case that happens.

My first doubt is about architecture. Is it viable to have a server build/project and a client one and use a local connection to make player-hosted servers and couch coop games? (Effectively having 2 unity instances in a machine)

Second, I'm not sure about going with photon bolt or forge remastered. We're not a big team so we may have to rely on Bolt's more out-of-the-box features, but I'm still in doubt.

Third, in case we go with Bolt AND we get to make dedicated servers, does anyone know about the new integration Photon announced with Zeuz? Or maybe you can recommend another provider?

Finally, whatever opinion and/or advice you may have will be very appreciated 😃

Thanks!

[last hornet]

is it predominantly a co-op experience @stray knoll ?

[stray knoll]

well in fact is a coop only experience

[last hornet]

@stray knoll i'm not sure bolt is a great fit? almost everything about it is optimized for competitive play

[vital hawk]

one huge downside of Bolt is it ships in precompiled DLL

[last hornet]

its basically a faithful adaptation of traditional source style shooter networking with hitbox rollback and whatnot for cheat prevention in competitive play, seems like a square peg in a round hole for a co-op dungeon crawler

and i say this as a big fan of bolt

[stray knoll]

Hmm... you have a point.

[last hornet]

I'd just go with PUN, especially if its only 2 players. dedicated servers seems excessive; PUN relaying is pretty ideal for your use case

[stray knoll]

I've been looking at both and I'm not quite sure about how they differ in the end. Their comparison page seems outdated as photon advertises bolt as server-first on some places and relayed p2p in others.

Do you have experience with both?

[last hornet]

@graceful zephyr could explain better, but i think bolt can be both. PUN is strictly relayed w/ MasterClient, but the way you do things is entirely different (Bolt events, replication etc. vs PUN RPC's, serialization and RaiseEvent)

[stray knoll]

And are those just different means to arrive at the same result, or do those differences get in the way of certain types of games?

In your opinion ofc

[last hornet]

well no matter how you use bolt it is aimed at arriving at a specific result (authoritative, valve style tick based networking where clients only forward input and the server verifies, resimulates, rewinds etc), PUN you are just given the tools to send /receive data and need to solve for everything from there

[stray knoll]

I get it better now. Thanks a lot!

[warm jungle]

Anyone here ever used https://github.com/RevenantX/LiteNetLib and can answer some nooby questions, mostly in regards to their NAT punch sample?

[weak plinth]

unity new networking instead of unet this is in alpha right?

[worn bramble]

unity new networking instead of unet this is in alpha right?
yeap

[weak plinth]

what can I use for production?

playfab

?

azure?

[graceful zephyr]

@weak plinth third party solutions

[compact bramble]

playfab and azure are random words that have no relation to unity networking

one of them is a backend service for tracking and holding data while the other is a generic cloud that does nothing

although I would like a unity version of a game backend so i dont need to go to third party

[graceful zephyr]

@compact bramble 😄

gonna have to wait for that lol

[compact bramble]

Yeah I guess lol

[graceful zephyr]

well there's 800 lines of C# on github if you want 😄

[compact bramble]

I'll take a look 😛

[weak plinth]

@weak plinth you could take a look at Photon and Forge Networking Remstered

the alpha unity stuff would require you to do all kinds of stuff yourself afaik, like reliability, bitpacking, compressing, security

[last hornet]

anyone ever have an issue in PUN where a match hosted in the Unity editor cannot be found by your builds, by either JoinRandomRoom or listing the rooms?

it just randomly started happening today :\

[celest hamlet]

Anyone knows if using UnityWebRequest with HTTPS encripts the communication?

[split crescent]

It does in the way that all SSL does

doesn't add any additional encryption ontop of that

but generally you're fine - only way to make it better is certificate pinning

[celest hamlet]

Okey, thanks!

[glass bobcat]

Is making your own networking lib a bad idea?

[vital hawk]

generally remaking any bigger library is a bad idea if you can by any means avoid it

but sometimes there simply isn't viable alternative that suits you and then you go down that rabbit hole

[glass bobcat]

what about experience wise?

[weak plinth]

i spent probably the last month looking in to making my own, trying things out, etc. wasted a lot of time and finally came to the realization that i would spend more than all the time i wanted to spend on the game making the networking library and in the end it would just be worse than something i could have used out of the box. imo, if you wanna make a game, use an existing lib. if you wanna make a networking lib, go ham, but you'll probably be doing that for a long time

[glass bobcat]

Aight

[compact bramble]

it's one of those if you have to ask then its a bad idea

[graceful zephyr]

@compact bramble lol

[fathom fern]

got a really odd bug happening. I have a node.js REST API hooked up to mongoDB via mongoose. when I send the JSON request via Postman it saves the new entry, but not when I send it by Unity. node.js logs that the new entry was saved successfully, but it's just not in the database...

anything obvious to check?

[weak plinth]

maybe fix the logging first and see where that takes you? sounds like whatever indicates that it's saved also has a bug

[fathom fern]

ugh, sorted it out. it was to do with coroutines executing out of sequence :p

[compact bramble]

shudders at the mention of coroutines

that's the thing, when you introduce coroutines in your code base, they can and will regularly stealthily sap perf and introduce little out of order bugs.. so convenient tho

I guess instead of focusing on making a new kind of allocation-free CR, Unity put their effort into ECS

[vital hawk]

people used to abuse the coroutines heavily on the Unity 4 times (I haven't really followed their current usage)

people literally reimplemented Update() with them and other weird stuff

I've mainly used them for some one shots or getting your code to run at certain point in the main loop (but now you can do it with custom playerloop too)

[stuck void]

I only use em for minor stuff, like when waiting for an async task to load

[glass bobcat]

you don't have to be precise in answering this question, but how much time would it take to create your own networking lib, for a first person multiplayer game using deterministic lockstep

just searching for an average

detailed answers are welcome ❤

[compact bramble]

It would take me about four years and I have over 30 years experience in general game programming, This is because the standard I would want it to run at would require me to specialise in it, and the testing alone would absorb at minimum, one of those years.

Intelligence wise I'm average, just quite competent and practical, so no genius here. But lots of practical experience.

TLDR please don't hurt yourself trying to make a networking library with that feature set

the problem here is the deterministic lockstep + FPS

And everything else that makes general purpose networking too... I can't imagine

[graceful zephyr]

@glass bobcat A first person game would not use deterministic lockstep for starters

@glass bobcat Creating a decent networking library from scratch, without prior experience, for a first person shooter type game... I'd say you'd need around two years maybe

It takes me about ~6 months to take a networking library from idea to usable beta, and I have around 8 years experience doing networking for games basically every day.

Bolt took me around 1.5-2 years of development time, it was released in 2014 and i started on what would become bolt 2012

It also massively depends on if you're making a generic library, or a library to network a specific game. The former is much more complicated.

You still need to know the same things for both a generic library or a specific game, but the amount of code and features you need to write is much less when networking a specific game.

[spring crane]

And by using something someone else made, you have the option to learn about things as you go/the hard way 😛

[glass bobcat]

I probably got my knowledge all wrong again great

Thanks for the answers! I'll probably go ahead and use an existing solution

@graceful zephyr if I were to make a "base" for a networking library and then add on to it to network a specific game, would that be faster?

[static maple]

I would suggest before making a networking library, first make a game or application that uses one? If you haven't already. Maybe make 3 or 4. Think of it like making a game engine.

[graceful zephyr]

@glass bobcat I dont know what a 'base' means

[glass bobcat]

nvm, just thought about it a little more, bad idea

thanks

[graceful zephyr]

There's generally 'three' pieces to a networked game

1. Transport (socket library)
2. Networking Higher Level API (i.e. Bolt/Forge/Mirror/UNET)
3. Actual integration into the game itself

You can of course build 2) in top of someone else 1)

Or just build 1) and then keep adding to 2) as you develop the game and need features, etc.

[glass bobcat]

aight

sounds good

Also @graceful zephyr if you don't use deterministic lockstep for a first person shooter, what do you use?

[graceful zephyr]

@glass bobcat two options generally, straight up delta snapshots (i.e. what counter-strike does) or something more flexible like eventual consistency (which is what halo:reach does)

[glass bobcat]

i knew about how CS does it, dunno what eventual consistency is

[graceful zephyr]

@glass bobcat unitys fps sample uses delta snapshots, so if you want to get a bid of a head start in your knowledge, dig into that code in detail over a few days

[glass bobcat]

👌

[graceful zephyr]

@glass bobcat for reference, deterministic lockstep doesn't work well for FPS games because they require you to have input delay on player input

rollback/predict can work for fps, but it's a lot more work than simple lockstep

[glass bobcat]

you still have the whole client prediction/server reconciliation thing right?

[graceful zephyr]

in rollback/predict? no