Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> I have tried using this API and found it incredibly technical and complex.

It is meant to be relatively low level. It's like WebGL, in order to draw a simple square on the canvas it takes likes 50 lines of code just to bind attributes to buffers and stuff, link, compile your shaders and execute the program on the GPU. It takes 3 lines with Three.js. You are probably not meant to use it directly if you don't know the domain, but use a library.

So I'd argue that if every single WebAPI was like that, the web would have been a failure.

At the same time, it's also a great playground if you INTEND to learn some things that can be useful OUTSIDE the browser. For instance, WebGL knowledge can be directly transferred to any OPENGL application, because the API is similar to C or C++ version of it.



I think the comparison to WebGL is apt. It's also worth adding that scaling WebRTC -- both to calls with more than a few participants, and to large numbers of users on a service -- is still a bit of a specialized undertaking.

Partly because it's intended to be a really, really flexible API, the abstractions are pretty leaky. You end up learning a lot about video codecs, UDP and RTP networking, and browser implementation details as you build out WebRTC applications!

(I'm the cofounder Daily.co of a YC company that tries to make WebRTC "just a few lines of code" for common use cases, and to be the best possible application-level APIs and global infrastructure for advanced developers.)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: