Return RakNet

Hello, I, like everyone else in this forum, is an ordinary developer ... As you already know, developers want to remove uNet which was not popular and in general I, like many, don’t understand why they reinvented the bicycle, in general I want to say that you did not delete RakNet from Unity and indeed why did you delete it? Can you answer? RakNet was the only network solution in Unity that made it possible to create a simple network game ... As you can see from the survey above, I suggest you return RakNet at least as a plug-in for the editor ... Just boiling ....

(Sorry, Google Translate)

 

It's unclear what's stopping you from writing the network engine on RakNet yourself

 

And what's the point? While you're going to write a wrapper in C # for RakNet, my great-grandchildren will have great-grandchildren, and Unity developers just need to make a small plugin that will allow you to use RakNet in Unity without problems ...

 

My money is on the legacy network was not going to work with ECS, which is also the reason stated for relatively quickly removing Unet.

Unity is not going to add RakNet back in. The time to fight for RakNet was several years ago. You might as well ask for a 32 bit editor and boo support while you are at it.

 

I take your money then.

ECS is compatible with any native networking transport, but for better performance N changes are required to the wrapper on top, and it works best with libraries where packets are abstracted as a completely independent unit of data (ENet for example). Regardless of ECS, in the case of UNet, it also requires to migrate UNet's reactor to Unity's Jobs system, but this is not that easy as it might sounds because the reactor is coarse-grained system while Jobs system is designed for fine-grained parallelism.

If it was my decision to replace UNet, then my arguments would be:

1. The transport is written in C++ while the vast majority of users here are not familiar with it, so making the source code open don't benefit them really. C# is the way to go with the wrapper of native sockets for better performance, lower latency, multiplexing, and so on.
   
2. It requires N changes to the UNet internal systems (and Unity's modules themselves) to make it work best in terms of performance with DOTS, not only ECS but the whole data-oriented stack in general.
   
3. Another chance to rehabilitate as a product that offers modern networking backend with rich feature-set and eventually return the credibility.