Hacker News Reader

Comment by bobajeff | Hacker News Reader

I'm surprised this is still being worked on I was under the impression that Google abandoned this.

Also, I would be interested to see a comparison to the wasm component model as it also seems to want to do the same things docker containers do.

View on HN

It never really got released so nobody is depending on it, so why would google abandon it yet?

They are running on Google's Nest Hub devices[0], so I guess this counts as a release?

[0] https://www.theverge.com/2021/8/18/22630245/google-fuchsia-o...

I think the lack of public information about their future plans for the project combined with the “killed by Google” meme got smashed together here and that is actually a really common perception but also one that is completely made up out of thin air.

It has been under heavy heavy development for many years now.

The fact that they are now starting to talk about it publicly now is probably a sign that they are looking to move beyond just IoT in the future.

For example, I know it’s coming to Android (not necessarily as a replacement but as a VM) and I know there is some plans around consolidating ChromeOS and Android as well. I expect that is also going to be another place we might see it before too long.

I know they are also working on a full Linux compatibility layer called Starnix [1] as well where I believe the goal is you can just run all your Linux workloads on Fuchsia without any changes is the goal AFAIK so you can probably extrapolate from there that the end state is going to be roughly in line with anywhere Linux runs currently is a good potential fit for Fuchsia and it will come with a lot of additional security guarantees on top of it that is going to make it particularly attractive.

[1] https://fuchsia.dev/fuchsia-src/concepts/components/v2/starn...

I think the problem is that "lack of public information about their future plans" is hard to differentiate from "no future plans exist". For a company that's in the past been known for their willingness to go for low-likelihood but high impact "moonshot" projects and seemingly open with some of their long-term plans (like everyone knowing that they were working on self-driving cars years before self-driving became a topic that people wouldn't be shocked to hear, like, their grandparents talking about), it's honestly pretty weird that the only device they've shipped with it was the Nest Hub, which apparently used to run on the same thing as a Chromcast (so which I don't even think was ChromeOS?).

If this were something that they were planning on using mainly for internal stuff, like for some sort of competitive advantage in data centers or something, I could understand the radio silence on future plans, but it's hard for me to imagine that's their main purpose when they're publicly putting it on stuff like the Nest Hub and Chromebooks (they didn't sell any with it afaik, but they published a guide for putting it on them). It really feels like they just don't know exactly what to do with it, and they're trying to figure that out as well. As for ChromeOS and Android, those already feel like a pretty good example of them not having a super clear initial product strategy for how they overlap (and more important, how they _don't_), so while having some sort of consolidation would make sense, it's not clear to me how Fuschia would help with that rather than just make things even murkier if they start pushing it more. I'd expect that consolidating them would start with the lower-level components rather than the UI, and my understanding is that Fuschia (as opposed to Zircon, which is the kernel) has quite a lot of UI-related stuff in it specifically with Flutter. I'm not saying you're wrong, since it sounds like you might have more relevant knowledge than me, but I can't help but wonder how much of this has really been planned in the long term rather than just been played by ear by those with decision-making power.

Fuchsia's UI layers are roughly equivalent to Wayland. There is a compositor, but no special alignment to any particular UI toolkit like flutter.

Fuchsia is not itself a consumer product, it's an open source project meant to be used to build a product. There is no application runtime for app developers to care about or UI for an end user to see. It would be strange to talk about things like mesa or the Linux kernel the way you are talking about fuchsia. There are software layers it does need to integrate with, but unless you work on those things, it's not really interesting to you.

Companies don't really discuss products they build using these open source building blocks while contributing to those projects until after the product launches either. It shouldn't really matter where and how it gets used to the end consumer, only that when it is used there are tangible benefits (more stable, less security problems, etc). I don't really understand why folks are so keen to understand what internal plans for using it may or may not be.

> Fuchsia is not itself a consumer product, it's an open source project meant to be used to build a product. There is no application runtime for app developers to care about or UI for an end user to see. It would be strange to talk about things like mesa or the Linux kernel the way you are talking about fuchsia.

The difference is that those aren't entirely funded and developed by a single a for-profit entity that presumably expects some sort of net positive results from the effort spent on them in the future. To be clear, I don't consider anything I said in my previous comment to be a reflection of whether Fuschia is useful or whether it has any technical merits; my commentary is intended to be entirely scoped to Google's _intent_ for Fuchsia, which is what I read the part of the top-level comment that the parent comment I responded to directly to be discussing.

It's certainly possible that you're correct that Google has had plans for Fuschia this whole time and didn't discuss them because they didn't think it was relevant, but I guess I just don't find that convincing enough to change my mind about what I perceive to be going on.

> It shouldn't really matter where and how it gets used to the end consumer, only that when it is used there are tangible benefits (more stable, less security problems, etc). I don't really understand why folks are so keen to understand what internal plans for using it may or may not be.

This is probably just a matter of differing personalities. I don't think I have any explanation for why I'm curious about the topic, but I just am. I think you could make any number of similar statements about not understanding why people find certain things interesting (sports, video games, celebrity gossip, etc.), and you wouldn't be wrong or right; in my experience, it's not a personal choice to decide to find something interesting or not.

It just feels really cynical to make everything about motivations and incentives. There are plenty of great projects that Google has produced that provide great external value without an interesting internal plan beyond using it to make products better. Golang, flutter, bazel, Gerrit, etc. I understand there are also examples where it uses it with some other intentions, such as the case of tensorflow or kubernetes, but I'm not really sure why one would think fuchsia is closer to those than it is to the former set.

It's certainly a bit cynical to view for-profit companies as being solely profit motivated, but I don't think it's _that_ cynical. To me, it almost seems a bit naive to assume that a company like Google would be doing something for purely altruistic motives in the absence of any evidence one way or another. I'm not a subscriber to Friedman's view that corporations are _required_ to maximize profits for their shareholders, but I do think that they all inevitably end up there in the absence of mitigating factors.

Google's goal is to make good products. That will enable making money. A positive externality of trying to create a good product is creating technologies such as fuchsia. I don't think there is more to it beyond that. If it's more complicated than that, that would probably be news to most people working on Fuchsia.

To me, one of the best features of ChromeOS is that it runs both Android and Linux. I have a number of telescopes that are controlled via Android apps, and being able to run astronomy processes apps like siril on the same platform is wonderful.

This is a potpourri of stuff over 7 years, sprinkled on a base of confirmation bias re: the common wisdom has a "perception...completely made up out of thin air", and a misunderstanding that speaking publicly in this individual talk represents a step change or something new.

I would bet, very, very, many dollars it is not coming to Android in any form, Starnix isn't coming soon if ever, and they're not looking to move beyond IoT. Long story short, it shipped on the Nest Hub, didn't get a great rep, and Nest Hubs haven't been touched in years because they're not exactly a profit center.

Meanwhile, observe Pixel Tablet release in smart display factor, Chrome OS being merged with Android, and the software-minded VP who championed the need for the project, moving on, replaced by the hardware VP.

When you mash all that together, what you get is: the future is Android. Or, there is no future. Depending on how you look at it.

(disclaimer: former Googler on Pixel team, all derivable from open source info. I wish it wasn't the case, but it is :/ https://arstechnica.com/gadgets/2023/01/big-layoffs-at-googl... https://9to5google.com/2023/07/25/google-abandons-assistant-... https://9to5google.com/2024/11/18/chrome-os-migrating-androi..., note 7d views on starnix bugs, all 1 or 0, with the exception of a 7 and 4 https://issues.fuchsia.dev/issues?q=status:open%20componenti...)

Personally, as an outsider, losing what ChromeOS was becoming (Google's spin on desktop Linux) is the saddest part of this for me.

God yes. It is/was so so so good. 2000s macOS values. I had to give it up when I left for a MacBook Pro*, and I still miss it.

* had to give it up? TL;DR: A key part of my workflow was being able to remote desktop into a Linux tower for heavy builds. Probably could have made it work anyway, obviously you wouldn't try building Android on a laptop, but a consumer app would be fine. I left to try and pick up some of the work I saw a lot of smart people do towards something better. And monetizing that in the short-term requires supporting iOS/macOS, which only compile on Mac

Fuchsia may not be outright dead, but it's definitely on life support and would've been killed a long time ago if senior people at Google weren't personally backing it. It had great foundations but without a concrete use case or product development was constantly pulled in different directions. It seemed like every year a new niche for Fuchsia was on the horizon, 6 months of development time would be dedicated to it, an extremely hacky demo would get the public hyped up, and then the whole thing would be abandoned because it didn't make any business sense. Starnix, for example, has been completely deprecated. There was even a newer system to replace it which also got cancelled.

* My knowledge is a couple years old at this point and I haven't kept up with recent developments so maybe the future is brighter than I think.

To wit, starnix has never been cancelled. Source: I work on fuchsia.

So the hope is that Starnix can emulate Linux syscalls well enough, while gVisor has been abandoned in later Google Cloud stuff because it couldn't emulate Linux syscalls well enough. Uh-huh.

Several operating systems successfully provide a Linux emulation mode. gvisor has very different constraints and requirements. It's also still heavily used and under active development so I'm not sure how you determined it is abandoned.

Newer generations of Google Cloud services run Linux VMs instead of gVisor.

I work on Starnix and I've never heard of anything meant to replace it. What are you talking about?

They might be thinking about POSIX Lite losing favor

Posix lite didn't lose favor. It's still an important part of writing fuchsia native software. It enables us to use the c++ and rust standard libraries with minimal upstream changes. It was never meant to enable running all existing programs, only lowering the barrier. There isn't really much software that has been ported to run on fuchsia natively. Instead runners are implemented or ported and those provide the environment applications require. For instance a flutter runner, web runner (chromium), and starnix (a Linux runner of sorts) provide the basis of running many existing applications.

Right, that is the current status

But the historical perspective is that Starnix is a relatively recent addition to Fuchsia. Even though Fuchsia is roughly 10 years old now, Starnix has only been useful for about 2 years (RFC 4 years ago)

Before Starnix came along to help run Linux apps, as you said, “There isn't really much software that has been ported to run on fuchsia natively”. Because POSIX Lite wasn’t / isn’t being used much. So I guessed the OP could have been thinking about that. But who knows.

There used to be patches in AOSP for Fuchsia but they all got reverted a couple years ago. I believe Starnix is the new strategy to get Android working on Fuchsia, if they are going to try for it.

> The fact that they are now starting to talk about it publicly now is probably a sign that they are looking to move beyond just IoT in the future.

You're reading too much into a conference presentation.

The team has been allowed to make conference presentations for many years, it's just that most folks haven't wanted to put in the personal effort. A few have in the past, one I know of was Petr: https://www.youtube.com/watch?v=DYaqzEbU0Vk

I think there's still significant investments into the project. Android is big enough so any sort of greenfield experiments cannot be done cheaply. Also there were so many custom variants of Linux based operating systems in Google and some of them moved onto Fuchsia. Probably it still have some values there.

Fuchsia has been on life support for a few years now, but not completely dead yet

There were 25 changes updated on the Fuchsia Gerrit in the last 15 minutes. It is much less dead than 99% of open source software projects.

The two biggest factors are how committed to this is Google and who is interested in developing and using this if Google drops it.

What do you think the answers to those are?

Factors for what? Deciding whether it's in life support? Deciding if it's okay for you to depend on it?

Are those 15 minutes representative? That seems pretty high if so.

During California working hours it might actually be a bit low honestly as you would see several revisions uploaded for code review per commit and there are on average 200-250 commits submitted daily to the primary fuchsia repo.

This is completely made up bullshit masquerading as a fact.

Reply on news.ycombinator.com