**Hello Emscripten** [GitHub](https://github.com/relgukxilef/hello-emscripten) · [Try](https://relgukxilef.github.io/hello-emscripten/) *Hello Emscripten* started as an experiment on abstraction. At the time I was working on a project at my company to add DirectX support to an OpenGL based renderer. The popular way of accomplishing this, as far as I have seen is to introduce an abstraction. Instead of using OpenGL or DirectX directly, you come up with your own API that wraps the underlying APIs. Unfortunately, comming up with an API is really hard. Many meetings and disagreements were had about what functions and what their signatures should be. And the parts of the API that we did end up implementing were difficult to use, because you couldn't just ask Google on how to use them. This frustration is what made me start to think about alternatives. I like to consider APIs their own thing. When you write a library, you're often writing both the API and the implementation. And both of those two things are difficult. So what if you only did one of them? What if you just took an existing API and used that as your abstraction? What if we just used an existing API and implemented it with OpenGL and DirectX? # VulkanGL At the time I had just started going through Vulkan tutorials, and I was thinking, Vulkan is not designed to be easy to use, but to be *easy to implement*. It offloads a lot of work to the user so the implementor doesn't have to think about it. Maybe I could implement a small subset of Vulkan in OpenGL? I started to write a simple Vulkan application that draws a triangle. You can link it with vulkan1 and it's an ordinary Vulkan application. Additionally, I wrote the library VulkanGL, an implementation of (a small subset of) vulkan.h in OpenGL. If you link the triangle application with VulkanGL instead of vulkan1, you get an OpenGL application with little overhead. And if you additionally link it with Emscripten's WebGL bindings, the triangles shows up in your browser. Exciting! *I want to write about design decisions in VulkanGL, like how command buffers are implemented, in the future* # Hello Also at the time, I was spending a lot of time with VRChat. VRChat is great, but I do have my gripes with it. I would like to see it facing more competition, and a killer feature I believe to be missing in VRChat is browser support. With WebXR and WebRTC, there should be no reason it shouldn't be possible to create a browser based social VR experience. The Japanese company HIKKY had already shared small social VR experiences running in the browser through Unity. Though I believe they are much more limited than VRChat and also have a lot less fidelity. I was convinced that, if one wanted to make such an application, it would have to use Emscripten, so why not try to turn my existing experiment into such an application? That said, I don't expect to be competing with VRChat any time soon. The whole thing is primarily an excuse to experiment. *I want to write more about Hello in the future*