Fuchsia 2024 Roadmap | Hacker News Reader

Fuchsia 2024 Roadmap

fuchsia.dev

44 pointskaycebasques25 comments

I worked on fuchsia for a while and so much of it is really interesting, and many of the people I was working with were incredibly smart, but it never really felt like Google as a whole really cared about it - there was the core kernel and OS that was designed sensibly and practically but a lot of the higher level parts of the system seemed to be used by people as a experimental playground at the cost of practical considerations.

It's a shame that Google & I were not a good match because I really enjoyed and valued the experience of working with the people on the teams I worked on/with.

> never really felt like Google as a whole really cared about it

I mean, their track record for products they initially cared about isn't great... so ya, it doesn't bode well for this one.

> Fuchsia to start providing a stable driver application binary interface (ABI) for at least six months for downstream products that use drivers compiled through the SDK. This confirms the platform's commitment to delivering reliable and long-lasting driver support.

6 months counts as "long-lasting" these days?

In Google Universe yes.

Wasn't this one of the key complaints with Linux that drove the creation of Fuchsia? Hopefully they're targeting more than 6 months eventually!

Yup, I thought so too. I mean 6 months is not even long enough to be development time for 1 major version of Android... They also cancelled plans to bring / maintain Chrome on Fuchsia this year. So not sure what their plans are with this OS anymore.

The team has been absolutely decimated and its essentially abandonware to Google at this point (i.e. runs on the Nest Hub ghosts ships). The long term story sold to the young and fresh-faced, you can kind of smell through the roadmap, RISC-V.

Baby steps. Committing to very long term horizons without sufficiently investing in your ability to uphold the necessary guarantees is challenging. Done poorly, we would either be unable to continue to evolve the platform or end up breaking things accidentally.

This is a system still in an active development phase, so I suppose they cannot commit to something like 5 years at this moment? Unlike API, ABI stability is a really hard thing to guarantee and it will haunt you to the end of eternity when you make a mistake.

Most of the roadmap is about Starnix. Starnix started serious development around late 2022 and is a way for Fuchsia to pretend to be Linux and run Linux programs without recompilation or modification. It is a pivot for the Fuchsia project, essentially giving up on the idea that most programs would be (re)written for Fuchsia

See https://fuchsia.dev/fuchsia-src/contribute/governance/rfcs/0...

It's 6 out of 18 bullets and arguably not all bullets are of equal size in terms of work. The starnix ones seem to be broken out to a more granular level.

I'm not sure it was ever a goal for all software to be written natively for fuchsia. Fuchsia only supports c++ in its sdk, which greatly limits the amount of software which can target it.

Starnix is clearly the focus now esp. if you focus on the platform goals and ignore the devex goals. Check out the previous roadmaps where Starnix was only mentioned once or not at all.

Aside: the 2024 roadmap also has way more devex goals than previous roadmaps, so it looks like devex in general is also a focus now

https://fuchsia.dev/fuchsia-src/contribute/roadmap

Reading anything about this project I always wonder if and how we ever get a completely new mainstream operating system.

I haven’t been keeping up, is Fuchsia in use much? From their site I see mention of Google nest hub as using it, but that is it.

No, almost nothing uses it.

I think they are struggling to find internal clients. It has on-boarded some customers using a customized Linux kernel but those are minor. ChromeOS was supposed to be the first big client but this effort is reportedly now deprioritized.

Any good use cases for Fuchsia over Linux?

The goal isn't to supplant Linux as a commodity kernel, but instead as a microkernel for integrated mobile devices. E.g., as a platform OS, not a general one.

So the short answer is "no, unless you use one of Google's hypothetical future post-Android devices".

Would Fuchsia supplant Android on mainstream smartphones or is it more intended for specialized devices like smart TVs and smart watches?

Any idea of that part was killed several years back, ensuing brain drain combined with firings makes it sort of a non-sequitor to even wonder. It is on life support already, and an obvious target for more Efficiencies™ should the 2020s turn rougher.

I'm not sure how you came to this conclusion, but that's not true. Well neither narratives are rather. There is no reason to believe fuchsia doesn't have a future anywhere Linux currently runs, but also, the goal isn't necessarily to supplant Linux.

Not if you need to run a mainstream web browser.

For those who are confused about this statement, like me, Google started bringing Chrome/Chromium support to Fuchsia a few years ago. But, earlier this year, they officially cancelled those plans: https://9to5google.com/2024/01/15/google-is-no-longer-bringi...

That really sounds like burying it.

Google probably decided to focus all on AI? Might make sense for them, but it is a shame, there seemed to be good engineering behind it.

It's worse than that - imagine if you had a team focused on a pie in the sky AI-based UI that ran on Fuchsia, but over-funded that, and it went nowhere, so it was killed, and then all of a sudden an LLM came out and all the miraculous UI parts seemed possible, but you were stuck with poorly run Android?

https://9to5google.com/2017/05/10/fuchsia-os-hands-on/

Reply on news.ycombinator.com