Hacker News new | ask | show | jobs
by specialist 655 days ago
IMHO, your strategy has always been the Correct Answer™. "Web browsers" should be based on URI/URL, HTTP, a canvas, and a stock virtual machine (eg WASM).

HTML/CSS support should just be another MIME type among many. Java Applets, Flash/Shockwave, PDF, Markdown, VRML-97, Quake .map, or whatever. Most sites would use one of the many stock renderer & runtimes. Some would implement their own, like your CSS reimplementation, as needed.

In other words, just skip the need for HTML shims.

Lastly, your efforts bring us that much closer to bringing us full circle. Next step: implement Wayland (and/or X11 if you're nostalgic).

1 comments

You're right, we feel trapped in the current model for serving content on the internet. We've talked internally about a browser API that would allow you to override certain CSS layout behavior. If I could extend / modify the browser's layout engine, I would gladly do so.
It’s a work in progress I thought, some of the APIs have shipped but it seems layout is behind flags in Chrome

https://ishoudinireadyyet.com/

https://github.com/w3c/css-houdini-drafts/blob/master/css-la...

Wow, this is really cool! Thanks for sharing.