Finally, a new start. 🎉
I started to work on ffmpeg.wasm around Oct. 2019, it was the time WebAssembly is still in very early stage and transpiling FFmpeg to ffmpeg.wasm was very challenging. (but possible as there was a pioneer project ffmpeg.js) Over these years, I have been trying to make ffmpeg.wasm to be production-grade, but failed as FFmpeg is such a powerful framework and WebAssembly is still evolving. It wasn't easy to support even one requirements (ex. RTSP) from the community and I found it really hard to continue sometimes. But with almost 11k stars, 600 forks, 4.1k used-by projects and 2.3M downloads in total, I still believe ffmpeg.wasm is a meaningful project to continue. That's why I decided to pause a little while and rethink what should be done in the next stage.
As the result, I started to work on ffmpeg.wasm 0.12.0, it is a major version and not backward compatible with 0.11.0. Massive issues are fixed in this version, including:
- Upgrade Emscripten to 3.1.40
- Upgrade FFmpeg to n5.1.3
- Docker build with cache (reduce time to build a new version)
- Support timeout(), terminate() and other fundamental operations
- Split libraries to multiple small libraries
- Rewrite the whole library with TypeScript
- Merge multiple repositories into one monorepo
- Offical Web Worker support
- Offical single thread and multi thread version support
- Enhanced documentation
- Official domain name
Lots of features are still in the backlog and it requires lots effort to even just compelete one of them, such as:
- Speed up ffmpeg.wasm using WebAssembly intrinsic
- Support RTSP protocol
- Support more popular libraries
I believe there is still a long way to go for ffmpeg.wasm to really become a production-grade library, but it is defintely a gamechanger and a chance to see the potential of WebAssembly and web technologies in general. That's why I am working on it and welcome to join us.
Hope you enjoy this release 😄
- Jerome