There’s Electron, but people complain that’s too memory intensive. Then there’s Tauri which seems nice, but it depends on WebView thus using a full browser.
What about a software app framework built from the ground up instead of bolting on top of an existing browser?
Not building a whole browser from scratch, an impossible task, but building just enough of a layout engine+run time to host web apps.
The goal here is, a dev writes an app for this framework using html, js, and css but it can also run in an existing browser. Because this framework would be so small and limited that it would be guaranteed to work somewhere greater.
The benefit being a lighter memory footprint because it’s no longer needing a full browser
solardev4 months ago
> Not building a whole browser from scratch, an impossible task, but building just enough of a layout engine+run time to host web apps.
Isn't that exactly what a webview IS...? It's been like that since the .CHM days (used for Windows help files, which were just web pages inside a lightweight web view).
What you're describing IS the layout engine, which Blink, Webkit, Gecko (and a bunch of deprecated older ones) already are. That is already the hardest part of a browser, which is why there are so many reskins of those three renderers.
If you want to build "just enough of a layout engine" from scratch, you pretty much ARE writing the hardest part of a browser.
HTML/JS/CSS have three decades of backward compatibility baggage using different nonstandard syntaxes and vaguely defined rules that makes any one of them very difficult to parse and optimize, much less the three together. It takes huge companies many years to come up with a new layout engine. Even Microsoft tried (many times) and gave up on that and went to Blink.
Maybe Ladybird will get somewhere, but I wouldn't count on it...
cranberryturkey4 months ago
primatejs now supports desktop