2 ms·
I want the exact opposite. All the current efforts in amateur browser building are all focused on documents. Gemini, and HN's other perennial favorite, Gopher,
by ampdepolymerase 6y ago
I want the exact opposite. All the current efforts in amateur browser building are all focused on documents. Gemini, and HN's other perennial favorite, Gopher, all look several decades out of date. I want a clean slate platform that is focused on high performance, superb design, and great developer productivity with an emphasis on 2021 hardware. A clean slate redesign of HTML with WebAssembly GUI widgets as a first class citizen (instead of the current "draw-to-canvas" hack). Gemini and Gopher will never gain traction if they perpetually look anachronistic. Forget about stuff like the semantic web. Those are dead ideas that will never take off unless they are properly backed by a strong business model. When I look at Gemini and the various new browser experiments, I get the feeling that HN web engineers lost their way. I know this forum worships Alan Kay and Brett Victor, but if all the new web browser projects here focus on the "good ole days" where everything is a semantically linked document, then there is no more room for growth. Where's the emphasis on UI and UX? Users clearly want apps that run fast and look pretty. For some reason engineers here reject that proposition and prefers ugly looking documents in the name of simplicity. Compiler and language technologies have advanced leaps and bounds since the Netscape days. We have .NET, LLVM, GraalVM, Julia, and two dozen different high speed production ready WebAssembly interpreters. 16+ cores hardware is starting to become the norm. Yet everybody is obsessed with the concept of a boring Tim Berners-Lee era document. Dream bigger. Figure out how to get native level performance on mobile devices if you are building a "web" browser. Make creating responsive GUIs as easy as traditional absolute positioned Windows Forms. Design a browser such that making a Stripe-level gorgeous landing pages becomes trival. Integrate Figma as a first class GUI design tool. Make collaboration and CRDTs first class (like the Braid RFC, https://braid.news https://braid.news). Use neural models to speed up content delivery and assets loading. The wave of unimpressive "new" browsers that cropped on HN in the last year reminds me of the Node.js web frameworks hype during the first half of last decade where there was a micro web framework on Show HN every week claiming to be the One Framework to Rule Them All despite mostly being a thin shim on top of express.js/the default node.js APIs and offering no real value. Dream bigger.
- oscargrouch 6y ago> WebAssembly GUI widgets as a first class citizen (instead of the current "draw-to-canvas" hack). Just a info. The 2d canvas of Chrome at least render to a texture layer only by appending commands to a display list that is only executed on the final stage (merged with all other commands) on the final stage in the gpu process. So it is as fast as it can get.. for instance its basically the same for the Chrome painted by C++ and you actually have to use canvas primitives to paint widgets. Once (if) WASM have access to something like this, it will be a canvas, the same one you have bindings for javascript. So i would not call it a hack, as you basically have raw access to paint primitives that are very optimized.