Yeah that's what immediately came to mind. This must be part of their effort to merge Chrome OS into Android. On the Chrome OS side they already said are going to be replacing the kennel and other system stuff with Androids guts.
That's sad news. ChromeOS is much faster and more efficient than Android. Turning off the Android subsystem in low-end Chromebooks is a huge performance boost, even when no Android apps are open.
To be fair, that’s likely because the Android subsystem is a virtual machine - not running multiple sets of system services / CPU emulation on a computer will make it faster pretty much universally.
It's not just the virtualization, ChromeOS has had a lot of work put into performance. The low-end ARM Chromebooks use the same hardware as budget Android devices, and they're noticeably faster. My Android phone uses more RAM doing nothing from a fresh boot than those Chromebooks even have.
That’s a good point actually. Windowed app performance on Android apps is especially bad since they’re not designed for it (most of my experience is on Android-x86, but window resizing performance is not great)
Fuchsia as the core makes much more sense. It replaces Linux for a start and completely changes the security model to something a LOT more defendable among a bunch of other benefits.
I worked on Android at Google until 2023 and can 99.999% confirm for you Fuchsia, as the outside understands it is DOA. (i.e. as some sort of next gen OS, and if not that, some kernel that's on track to replace Linux in Android)
Long story short is you can imagine in 2019 there was X amount of engineers, 95% on Android and 5% on what you'd call Fuchsia.
The central argument up top became about why the renegade band that split off from Android/Chrome etc. to do Fuchsia in...early to mid 2010s?...and if it was going to provide a significant step forward. This became framed in terms of "$ of devices shipped", in which case, there was no contest.
Funnily enough this very article is about N dominos down from there (de-investment in Fuchsia, defenestration of head software guy of Android/Chrome/Chrome OS etc., ex. Moto hardware guy is in charge now)
Don't read this comment too closely, I was not in the room. For example, I have absolutely no actual quotes, or relayed quotes, to 100% confirm some set of individuals became focused on # of devices shipped.
Just obsessive enough to track wtf was going on, and on big enough projects, and trustworthy and hard working, and clearly without party or clique, such that I could get good info when asked, as it was clear my only concern was making things that were good and making sure all of Google's products could be part of that story.
Thanks for that insight. Obviously there’s a lot of context in there that isn’t at all clear outside.
One part I find hard to reconcile with all of that is that even just looking at public facing stuff alone it seems to be under VERY active development.
You're seeing the public facing side of it as a wholesale Android replacement, think of this repo as the tip of the iceberg: well, it used to be, but the rest of the iceberg disappeared.
It's still shipping, successfully, on millions(?) of devices yearly via Nest Hubs and such. Never say never. But all public-facing signs are fairly clear, the big move is ChromeOS into Android, not Android onto Fuchsia - and note how invested Google is in the Nest Hubs (read: not at all, languished for years now)
(I'm also curious if we could get a stats-based thing on, say, 2019 vs. 2024, My out-of-thin-air prior would be 30% more activity in 2019 than now. But I also figured there was 10 commits a day now, not 100.)
(I tried checking out the whole repo, but then all the apps on my macbook informed me en masse there was 0 disk space left :X Doesn't look like there's a GitHub mirror)
(cheers btw, you're my kind of people, one of the more soul-sucking parts of Google was finding out that kind of person is few, and far between) (i.e. curious and into It, not just here to make your boss or partner happy)
I guess that’s the thing, when I now look at the commits it’s way past what’s reasonable for nest. I see chromium is rolled back in for example, flutter is back in there, everything seems to be getting the bazel treatment, I see a bunch of non publicly announced releases this year… like things are clearly happening on some level in a way that’s not entirely consistent with the idea of a product that is in the midst of a death roll.
Do you think it’s possible that thinking evolved slightly beyond the number of devices shipped in the meantime perhaps?
I’d always been on the opinion of even just assuming they got to a point where the whole starnix linux syscall compatibility layer thing (which is what I see a lot of recent commits pointing towards) to a good point and stopped that it would still make sense in so many use cases that it would have been justified.
It would also change the number of devices shipped question from 1 dying product line to billions overnight which certainly might change the conversation and furthermore would put them into a pretty unique commercial position for more high assurance computing given the security model which would have a lot of follow on effects presumably
Also thanks again for taking the time to answer, you’re right I am genuinely very curious about this.
