2,029 karma · joined July 18, 2011
Here was the announcement:
https://v8.dev/blog/code-caching-for-devs
https://github.com/thlorenz/v8-perf/blob/master/snapshots%2B...
TikTok aggressively funnels you into the native app for various business reasons, but the experience is well polished if they'd let you use it all the time.
Safari (really that iOS won't let you pick a different browser) has definitely been holding back the growth of PWAs though. Without decent install to home screen support (apple doesn't implement install banners for web), no push notifications, many of the native API integrations missing, it's hard to go too far down the web path before the business folks want the native app for growth angles. It's a lot more about making money than the better UX that can be had with native.
Ex. If you can't run the Android or iOS YouTube app, what you have left is the web one... which could be amazing, but is hardly a fully functional PWA because Google would rather you used the native app.
Firefox OS would have failed all the same had it been based on Qt, or even if it was based on Java without the Android runtime and SDK.
https://www.nytimes.com/2022/12/29/technology/twitter-elon-m...
What's the speed enhancements?
https://en.m.wikipedia.org/wiki/LCARS
LCARS has always seemed so beautiful and simple, but in practice you can't get real UI complexity into a button layout like that. On the shows the buttons mostly display random numbers (which of course made it easier to produce and reuse the panels).
Also even with seizures you can drive if you're on medication and haven't had one in a while.
https://floooh.github.io/sokol-html5/
Interestingly the imgui and nuclear both work on mobile while the microui demo doesn't appear to handle any input. Imgui works surprisingly well (although everything is a bit small for touch).
That said Projector round tripped all the key strokes to get new draw commands (unlike VSCode) which also results in lag. The new IntelliJ remote architecture is much better, and it seems Uber is moving that way too.
That architecture has a major flaw though. A default install was fast, but a real configuration with all the plugins ended up quite slow.
VSCode's architecture worked much better in real world configurations, and it turns out better performance wins in the editor space, even if it meant losing certain customizations (up to a limit). The architecture also allowed seamless remote editing, something Atom never could have done.
The interesting thing has always been that both were built on Electron. Atom's developers had built all the tools and primitives needed, they just didn't have the understanding of the long term ecosystem effects of their design.
All those legacy features and existing pages are the back pressure that keeps the evolution of the web stable.
You can also set the "temperature" parameter to 0 to remove the non determinism (though I'm not sure that's exposed in the web interface).
If you pay a premium price you expect premium service.
There's no competition for airplane wifi though, you either use it or you don't. It's not like you choose between two different vendors... So they're competing against no wifi at all which allows for pretty mediocre and unreliable service as that's "better than nothing".
These free ML models exist in a slightly different world.
It's also possible you're in an A/B test to show a different ad volume.
In general sample size of 1 for ads (or feed ranking) is hard to reason about.
It's a pretty cool project, and the source is a fun way to learn about web standards (same with reading Ladybird browser code). The major browsers are so complex they're not as approachable.