about VCS in server rendering env

[chenzx]

I'm very curious about VCS impl internals in server rendering env after reading daily blog articles, now i see, it's just rendering React JSX components to canvas2d (so display list just mapped to ctx.drawXxx commands?)

I've built a C++ native 2D rendering engine to canvas2d and find some limits: canvas2d has no 3d perspective transform and also no fine-grained text font rendering contol. So i later use Wasm+WebGL instead of canvas2d.

Chrome's web apis support recording element to video stream and use it in webrtc, all hardware-accelerated. But i know chrome headless (or use electron) in linux may have no full 0-copy GPU compositing support...

I'd like to see a native C++ impl of VCS(by customizing chromium?) because JSX is elegant language for app/video-overlay UI description (better than Flutter, or Kotlin Compose, or Qt QML), which will be used in both embedded native (like IVI) and cloud-rendering scenarios.

[chenzx]

Recently React Native community has a "Static Hermes" try, which in the future may lead to a full-featured "Statically compile HTML fontend apps to native apps"