Jump to content

[1.6.2]Separating the code on client and server sides?


Naiten

Recommended Posts

So, i have an entity with tons of different code in it's onUpdate(). And i know i can use if(worldObj.isRemote) to check if the code is run on server or client side. Now i want to understand which parts should i run on every side. For example, which side should i use when operating motion variables or moveEntity()?

If i helped you, don't forget pressing "Thank You" button. Thanks for your time.

Link to comment
Share on other sites

Well, I would like to think that would be simple logic problem, but oh well... ( I apologise if this is taken insultingly, I am not intending it to be that way. )

 

I also apologise if this seems extremely childish to you, and seems insulting. I am not trying to be.

 

 

 

Think about this logically. What is a client? A client takes the data from the server, and displays that data in some way. The server handles the data, and sends it to the client to be displayed.

 

So, going by what I have just said ( if it was unclear, please say so, I am still trying to get the hang of explaining things and your input would be helpful :) ), anything that deals with visuals, so your entity movement and stuff like that, is done client side. The server handles the information like the placement of the entity on the map, the locations of blocks etc. Once again, the server handles the data, the client displays the data.

 

I hope that helps :) If it doesn't, I am sorry.

I am Mew. The Legendary Psychic. I behave oddly and am always playing practical jokes.

 

I have also found that I really love making extremely long and extremely but sometimes not so descriptive variables. Sort of like what I just did there xD

Link to comment
Share on other sites

I can't get why server would store position, but not motion. And how is movement or collision (physics, in more general terms) dealing with visuals? And if every client process movements by itself, can't they (clients) run with different frequency and desync?

If i helped you, don't forget pressing "Thank You" button. Thanks for your time.

Link to comment
Share on other sites

I can't get why server would store position, but not motion. And how is movement or collision (physics, in more general terms) dealing with visuals? And if every client process movements by itself, can't they (clients) run with different frequency and desync?

 

Like I said.. It has flaws. Thanks for pointing that out. And I did leave room for expansion in my explanation didn't I? So that means the rotation could be kept on the server, but I doubt that myself.

I am Mew. The Legendary Psychic. I behave oddly and am always playing practical jokes.

 

I have also found that I really love making extremely long and extremely but sometimes not so descriptive variables. Sort of like what I just did there xD

Link to comment
Share on other sites

It doesn't really matter if clients don't see exactly the same entity animation at the same time.

Just look at the vanilla "flying" player animation to see how senseless animations are.

What matters is collision and positions, which are handled by the server.

 

As for the question at hand, it is not needed to separate client and server sides in most cases.

Of course, client-side only methods (rendering, animation...) should only be called client-side.

And if there is any chance of desync between client and server (say, you add a bit of random in the logic), stay on server side.

This is what i would say in general.

Link to comment
Share on other sites

In my vehicle implementation, my onClientUpdate is pretty small -- it just applies any motion that it had received from the server.  Thats it--little/no calculation, no collision checks, just blindly/dumbly apply what the server sent.  My server-side calculates motion, applies collision, applies rotation, applies user input, and sends update-packets to client side with the intended positioning.  Client side takes that information and figures out how to be where the server says it should be within the allotted time (3-5 ticks generally) (very similar to the vanilla boat rotation-delta updating methods). 

 

How much needs to be done client side will mostly depend on how you are calculating motion, and where you are doing collision checking at.  At the minimum-- you will need to update any movement/animation variables -- e.g. leg position, wheel rotation, as well as applying any motion you had calculated (you can techincally do motion calculations client side as well and use position synch packets to keep things lined up...but I figured...why calculate something twice?).

 

Mew has the correct approach though -- think exactly what is needed client side?  You need to know where it is at, how it is moving (both how it is rotating and moving in the world), possibly the health value (for health bars), and any other information needed for client-rendering.  Pretty much everything else can be server-side only.

 

(I can provide links to source examples if needed)

Link to comment
Share on other sites

In my vehicle implementation, my onClientUpdate is pretty small -- it just applies any motion that it had received from the server.  Thats it--little/no calculation, no collision checks, just blindly/dumbly apply what the server sent.  My server-side calculates motion, applies collision, applies rotation, applies user input, and sends update-packets to client side with the intended positioning.  Client side takes that information and figures out how to be where the server says it should be within the allotted time (3-5 ticks generally) (very similar to the vanilla boat rotation-delta updating methods). 

 

How much needs to be done client side will mostly depend on how you are calculating motion, and where you are doing collision checking at.  At the minimum-- you will need to update any movement/animation variables -- e.g. leg position, wheel rotation, as well as applying any motion you had calculated (you can techincally do motion calculations client side as well and use position synch packets to keep things lined up...but I figured...why calculate something twice?).

 

Mew has the correct approach though -- think exactly what is needed client side?  You need to know where it is at, how it is moving (both how it is rotating and moving in the world), possibly the health value (for health bars), and any other information needed for client-rendering.  Pretty much everything else can be server-side only.

 

(I can provide links to source examples if needed)

 

Thanks for the support :D But I would be inclined to follow GotoLink on this, why would you calculate motion/rotation on the server? That is just extra resources used on the server and it does't really matter if the clients are out of sync... Does it? Please correct me if I am wrong in this!

I am Mew. The Legendary Psychic. I behave oddly and am always playing practical jokes.

 

I have also found that I really love making extremely long and extremely but sometimes not so descriptive variables. Sort of like what I just did there xD

Link to comment
Share on other sites

Well, for my scenario, ALL of my motion is calculated server side.  It needs to calculate rotation because the forward motion/direction depends on the rotation.  You don't necessarily need to calc rotation server side (can interpolate from the motion vector on the client side), but in my case, my vehicles have forward and reverse, which necessitated using rotation in the basic motion calculations, and synching this rotation to clients (they don't calculate based on it, merely display whatever the server tells them)

 

As to the other...how do you _not_ calculate motion server side?  If the entity on the server doesn't calculate motion (and move), the client side entities certainly won't move (or will be out of position client side, and then collision detection wont work properly).

 

Essentially I moved over to a completely server-authenticated/calculated vehicle movement system, though it is probably overkill for most peoples purposes (I required smooth and (mostly) synched motion between server/clients, and that is the most reliable way I found to achieve it).

 

Anyway, I'm not meaning to argue on it, there are lots of ways to handle motion/synching depending upon your needs (dead reckoning, predictive, checked, lots of others) -- I was merely wanting to point out that you can indeed separate the code for client/server, for both easier reading of code, and possibly increased performance -- after all, why calculate something twice (or more, in the case of multiple clients), when once will do.

 

 

Link to comment
Share on other sites

Well, for my scenario, ALL of my motion is calculated server side.  It needs to calculate rotation because the forward motion/direction depends on the rotation.  You don't necessarily need to calc rotation server side (can interpolate from the motion vector on the client side), but in my case, my vehicles have forward and reverse, which necessitated using rotation in the basic motion calculations, and synching this rotation to clients (they don't calculate based on it, merely display whatever the server tells them)

 

As to the other...how do you _not_ calculate motion server side?  If the entity on the server doesn't calculate motion (and move), the client side entities certainly won't move (or will be out of position client side, and then collision detection wont work properly).

 

Essentially I moved over to a completely server-authenticated/calculated vehicle movement system, though it is probably overkill for most peoples purposes (I required smooth and (mostly) synched motion between server/clients, and that is the most reliable way I found to achieve it).

 

Anyway, I'm not meaning to argue on it, there are lots of ways to handle motion/synching depending upon your needs (dead reckoning, predictive, checked, lots of others) -- I was merely wanting to point out that you can indeed separate the code for client/server, for both easier reading of code, and possibly increased performance -- after all, why calculate something twice (or more, in the case of multiple clients), when once will do.

 

I see your point.. That makes sense now xD Thanks for clearing that up. As I also had said, please point out where I am wrong :D And you did. So thank you for extending my knowledge in this area :)

I am Mew. The Legendary Psychic. I behave oddly and am always playing practical jokes.

 

I have also found that I really love making extremely long and extremely but sometimes not so descriptive variables. Sort of like what I just did there xD

Link to comment
Share on other sites

Well, this is what i have come to:

    public void onEntityUpdate(){
        /* 
         * Code that is not concerned to motion
         * <...>
         */
        if(this.worldObj.isRemote){
            this.setPositionAndRotation(this.posX, this.posY, this.posZ, this.rotationYaw, this.rotationPitch);
            this.moveEntity(this.motionX, this.motionY, this.motionZ);
            this.correctCartPositionOnRail();
            this.motionX *= 0.95;
            this.motionY *= 0.95;
            this.motionZ *= 0.95;
        }else{
            /*
             * Code that does something with motion
             * <...>
             */
            this.moveEntity(this.motionX, this.motionY, this.motionZ);
            this.correctCartPositionOnRail();
            this.motionX *= 0.95;
            this.motionY *= 0.95;
            this.motionZ *= 0.95;
        }
    }

And, what i got:

http://www.youtube.com/watch?v=WA5JJtbYoNc

Not really bad, but still kinda desynced, as i see.

If i helped you, don't forget pressing "Thank You" button. Thanks for your time.

Link to comment
Share on other sites

Damn. I lost the sight of that.

Tried what you've said, and the engine wasn't glitching. But wheel rotation broke, so i started messing with the code, and now i can't get «non-glitchy motion» state again -.- Facepalm.

If i helped you, don't forget pressing "Thank You" button. Thanks for your time.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Unfortunately, your content contains terms that we do not allow. Please edit your content to remove the highlighted words below.
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.



  • Recently Browsing

    • No registered users viewing this page.
  • Posts

    • rp.crazyheal.xyz mods  
    • I'm developing a dimension, but it's kinda resource intensive so some times during player teleporting it lags behind making the player phase down into the void, so im trying to implement some kind of pregeneration to force the game loading a small set of chunks in the are the player will teleport to. Some of the things i've tried like using ServerLevel and ServerChunkCache methods like getChunk() dont actually trigger chunk generation if the chunk isn't already on persistent storage (already generated) or placing tickets, but that doesn't work either. Ideally i should be able to check when the task has ended too. I've peeked around some pregen engines, but they're too complex for my current understanding of the system of which I have just a basic understanding (how ServerLevel ,ServerChunkCache  and ChunkMap work) of. Any tips or other classes I should be looking into to understand how to do this correctly?
    • https://mclo.gs/4UC49Ao
    • Way back in the Forge 1.17 days, work started for adding JPMS (Java Platform Module Support) to ModLauncher and ForgeModLoader. This has been used internally by Forge and some libraries for a while now, but mods (those with mods.toml specifically) have not been able to take advantage of it. As of Forge 1.21.1 and 1.21.3, this is now possible!   What is JPMS and what does it mean for modders? JPMS is the Java Platform Module System, introduced in Java 9. It allows you to define modules, which are collections of packages and resources that can be exported or hidden from other modules. This allows for much more fine-tuned control over visibility, cleaner syntax for service declarations and support for sealed types across packages. For example, you might have a mod with a module called `com.example.mod` that exports `com.example.mod.api` and `com.example.mod.impl` to other mods, but hides `com.example.mod.internal` from them. This would allow you to have a clean API for other mods to use, while keeping your internal implementation details hidden from IDE hints, helping prevent accidental usage of internals that might break without prior notice. This is particularly useful if you'd like to use public records with module-private constructors or partially module-private record components, as you can create a sealed interface that only your record implements, having the interface be exported and the record hidden. It's also nice for declaring and using services, as you'll get compile-time errors from the Java compiler for typos and the like, rather than deferring to runtime errors. In more advanced cases, you can also have public methods that are only accessible to specific other modules -- handy if you want internal interactions between multiple of your own mods.   How do I bypass it? We understand there may be drama in implementing a system that prevents mods from accessing each other's internals when necessary (like when a mod is abandoned or you need to fix a compat issue) -- after all, we are already modding a game that doesn't have explicit support for Java mods yet. We have already thought of this and are offering APIs from day one to selectively bypass module restrictions. Let me be clear: Forge mods are not required to use JPMS. If you don't want to use it, you don't have to. The default behaviour is to have fully open, fully exported automatic modules. In Java, you can use the `Add-Opens` and `Add-Exports` manifest attributes to selectively bypass module restrictions of other mods at launch time, and we've added explicit support for these when loading your Forge mods. At compile-time, you can use existing solutions such as the extra-java-module-info Gradle plugin to deal with non-modular dependencies and add extra opens and exports to other modules. Here's an example on how to make the internal package `com.example.examplemod.internal` open to your mod in your build.gradle: tasks.named('jar', Jar) { manifest { attributes([ 'Add-Opens' : 'com.example.examplemod/com.example.examplemod.internal' 'Specification-Title' : mod_id, 'Specification-Vendor' : mod_authors // (...) ]) } } With the above in your mod's jar manifest, you can now reflectively access the classes inside that internal package. Multiple entries are separated with a space, as per Java's official spec. You can also use Add-Exports to directly call without reflection, however you'd need to use the Gradle plugin mentioned earlier to be able to compile. The syntax for Add-Exports is the same as Add-Opens, and instructions for the compile-time step with the Gradle plugin are detailed later in this post. Remember to prefer the opens and exports keywords inside module-info.java for sources you control. The Add-Opens/Add-Exports attributes are only intended for forcing open other mods.   What else is new with module support? Previously, the runtime module name was always forced to the first mod ID in your `mods.toml` file and all packages were forced fully open and exported. Module names are now distinguished from mod IDs, meaning the module name in your module-info.java can be different from the mod ID in your `mods.toml`. This allows you to have a more descriptive module name that doesn't have to be the same as your mod ID, however we strongly recommend including your mod ID as part of your module name to aid troubleshooting. The `Automatic-Module-Name` manifest attribute is now also honoured, allowing you to specify a module name for your mod without needing to create a `module-info.java` file. This is particularly useful for mods that don't care about JPMS features but want to have a more descriptive module name and easier integration with other mods that do use JPMS.   How do I use it? The first step is to create a `module-info.java` file in your mod's source directory. This file should be in the same package as your main mod class, and should look something like this: open module com.example.examplemod { requires net.minecraftforge.eventbus; requires net.minecraftforge.fmlcore; requires net.minecraftforge.forge; requires net.minecraftforge.javafmlmod; requires net.minecraftforge.mergetool.api; requires org.slf4j; requires logging; } For now, we're leaving the whole module open to reflection, which is a good starting point. When we know we want to close something off, we can remove the open modifier from the module and open or export individual packages instead. Remember that you need to be open to Forge (module name net.minecraftforge.forge), otherwise it can't call your mod's constructor. Next is fixing modules in Gradle. While Forge and Java support modules properly, Gradle does not put automatic modules on the module path by default, meaning that the logging module (from com.mojang:logging) is not found. To fix this, add the Gradle plugin and add a compile-time module definition for that Mojang library: plugins { // (...) id 'org.gradlex.extra-java-module-info' version "1.9" } // (...) extraJavaModuleInfo { failOnMissingModuleInfo = false automaticModule("com.mojang:logging", "logging") } The automatic module override specified in your build.gradle should match the runtime one to avoid errors. You can do the same for any library or mod dependency that is missing either a module-info or explicit Automatic-Module-Name, however be aware that you may need to update your mod once said library adds one. That's all you need to get started with module support in your mods. You can learn more about modules and how to use them at dev.java.
    • Faire la mise à jour grâce à ce lien m'a aider personnellement, merci à @Paint_Ninja. https://www.amd.com/en/support 
  • Topics

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.