Re:Runs on the Librem 5 Smartphone – Week 2
by hajile on Friday July 12, 2019 @03:57PM
(#20423915)
I think Firefox OS didn't have to fail.
1. Allow non-kernel updates without carrier interference. Google has had to learn this lesson the hard way. Let the carriers lock down the lower layers, but Firefox and the user-facing stuff should still be able to update well enough for a few years.
2. Don't launch on crap, low-memory devices. KaliOS works on low-end devices, but that's because it is hyper particular about which apps run and what web APIs they use (and a lot of websites don't run very well). There's no reason a web device can't work for high-end devices. Most people run the same 50 apps as everyone else (and 25 of those ship with the phone anyway). High-end devices also open up other options like running Android in a linux container (like ChromeOS does).
3. Make WebAssembly (then asm.js) first-class. "Native" wasm apps with more direct access to APIs and access to a standard set of canvas widgets (sorry, HTML sucks for a lot of things). New APIs need to be available for more specific cores (especially things like DSP or ML units). Finally, add support for fully-native processes for trusted programs because sometimes 25-50% slower is too much and sometimes a spec (eg, Vulkan) simply isn't available as a webAPI yet but competition is needed here and now.
4. Be different. Make something that tries to clone Android and people will expect it to be Android. Firefox OS would have gone much farther looking and acting like webOS (which is still a better UI than Android more than a decade later). You can't disrupt a free, mature project with an identical, but immature project and a bunch of ideals that 95% of people simply don't understand.
1. Allow non-kernel updates without carrier interference. Google has had to learn this lesson the hard way. Let the carriers lock down the lower layers, but Firefox and the user-facing stuff should still be able to update well enough for a few years.
2. Don't launch on crap, low-memory devices. KaliOS works on low-end devices, but that's because it is hyper particular about which apps run and what web APIs they use (and a lot of websites don't run very well). There's no reason a web device can't work for high-end devices. Most people run the same 50 apps as everyone else (and 25 of those ship with the phone anyway). High-end devices also open up other options like running Android in a linux container (like ChromeOS does).
3. Make WebAssembly (then asm.js) first-class. "Native" wasm apps with more direct access to APIs and access to a standard set of canvas widgets (sorry, HTML sucks for a lot of things). New APIs need to be available for more specific cores (especially things like DSP or ML units). Finally, add support for fully-native processes for trusted programs because sometimes 25-50% slower is too much and sometimes a spec (eg, Vulkan) simply isn't available as a webAPI yet but competition is needed here and now.
4. Be different. Make something that tries to clone Android and people will expect it to be Android. Firefox OS would have gone much farther looking and acting like webOS (which is still a better UI than Android more than a decade later). You can't disrupt a free, mature project with an identical, but immature project and a bunch of ideals that 95% of people simply don't understand.