Fuchsia F14 | Hacker News Reader

Fuchsia F14

fuchsia.dev

58 pointsmbStavola56 comments

Interesting that F13 and F14 (this release) aren't listed on the parent page: https://fuchsia.dev/whats-new/release-notes

Regardless, from a "software tourism" perspective, I'm happy to see Google is still putting work into Fuschia. I read somewhere that the team has been really impacted by layoffs.

I personally think the time is ripe for an OS with more modern security primitives than Linux/Windows.

It would be nice if macOS had some competition.

No joke, this was my main hope for fuschia. I don’t think Linux or the BSDs can compete without a total re-think of their multimedia and GUI stacks (and the needed re-think does not look like Wayland).

I’m no longer too hopeful that Fuscia will be it, with its snail’s-pace development and seemingly having been back-burnered on top of that, but it’s what I’d hoped to get out of it.

Wayland is close to macOS windowing, just like systemd is close to launchd. That’s how open source can approach the competence of macOS, or Windows for that matter.

> the competence of macOS, or Windows

I hope that open source stays as far as possible from "the competence of macOS, or Windows".

Windows is a PITA to use.

And yet it can render flawless frames and handle background services, just like macOS. The policy choices are sometimes unfortunate, for sure.

As someone that uses Linux on my home computer home and has to use MacOS at work, I actually laughed at this comment.

MacOS is awful. I would rather use Windows than this crap.

I also use Linux for personal computers and have a company-issued MacBook. I definitely prefer Linux over MacOS but you couldn’t pay me to use Windows.

Windows 10 with WSL was at least serviceable for the most part, and is useful for gaming at least.

Make no mistake, I have no love left for Windows. Discussing Windows vs MacOS is like discussing which piece of turd smells worse.

To be fair, if the turd is from different animals the smell of one might be wildly stronger than the other.

But yeah, I'm on linux too. Although I would love to have a distro with main support for kde.

What are the things that you find awful about MacOS?

- Commandline tools are gratuitously a little bit different.

- I never managed to find an usable keyboard shortcut for some GUI operations.

- Things that worked stopped working after an update because of security.

A modern OS is fine so long as it's not led by Google.

> I personally think the time is ripe for an OS with more modern security primitives than Linux/Windows.

What's currently a modern OS that allows for zero security primitives in favor of faster everything?

FreeRTOS

For people who understand this type of things, how would you describe the internals of this OS compared to Linux? How is this better/worse than Linux from a code standpoint?

How does HN feel about every tech submission having a quickly topvoted thread that is either an explainer or asking for an explainer?

It's kind of interesting to me to see the various takes. But I also kind of feel like folks should just go search for the thing & find & recommend something, rather than asking what is this for each time (sometimes very rudely! Not here thanks you!). https://hn.algolia.com/?q=fuchsia

The only real difference seems to be in the kernel. And that one difference is that it's "capability-based", which in practice means that programs are given keys by the kernel, which they can use later to make syscalls. This theoretically makes security easier to reason about.

There are tons of other differences and I'm not going to bother to list them all out here, but this is just not accurate.

How is this comment helpful if you won't list the differences or explain why it's not accurate?

I find it helpful.

"Warning: don't trust this"

Could you bother linking to a place listing them all out there?

I think this is missing the forest for the trees: most of what would normally be a systemcall to the kernel in Linux is instead a capability-based IPC to other userspace processes with limited privilege. This includes device drivers, the networking stack, filesystems, etc.

If it was a company you could trust with LTS, it would be interesting.

Seven years of Pixel updates and 10 years of ChromeOS updates aren't nothing.

Its also completely and totally unproven. I had chrome laptops that warned that they were out of support two years ago, less then a year after I bought it. The whole process disguested me so much that I replaced them with macs.

lets see google actually deliver.

It's pretty much nothing, and the only reason anyone thinks it's impressive is because the competition is so abysmally bad.

We've got free software projects like the BSDs and Linux supporting hardware that's decades old with their latest releases, and we should be impressed when giant companies with infinitely more resources like Google and Apple can't even stretch OS support to a mere decade?

In different areas, though. Windows also supports some very old hardware and the FOSS community is also out of luck when phone chip makers stop releasing blobs/kernels.

Google is in the best position to tell chipmakers they can't use proprietary blobs. They just aren't motivated enough to try.

Beside Google, is there other companies contributing to and building products with Fuchsia?

The ones I could find:

Samsung: https://fuchsia-review.googlesource.com/q/owner:sukka.kim@sa... https://fuchsia-review.googlesource.com/q/owner:j_yoon.choi@...

Amlogic: https://fuchsia-review.googlesource.com/q/owner:dingxin.jin@... https://fuchsia-review.googlesource.com/q/owner:frank.chen@a... https://fuchsia-review.googlesource.com/q/owner:manliang.tan...

Linaro: https://fuchsia-review.googlesource.com/q/owner:jagdish.gedi...

Samsung is porting their f2fs filesystem and doing some stuff in the storage stack. The other 2 companies are doing device drivers.

what's this for again?

So google can inject adware into the kernel itself instead for Android.

Yeah Google didn't control enough of the stack for their liking.

Is it still being funded by Google? That seemed to be in doubt earlier in the year.

It's an open source project that is developed by Google(not funded by). It's used on their Google hub smart devices. They had layoff in the department but until they decide to discontinue their smart devices then they will continue active development of this.

Is it necessarily tied to smart devices? I always wondered if it was also partially desired by Google for displacing Linux in some roles inside their data centers. In any case, they are still landing hundreds of changes per day, so it doesn't even resemble a dead project.

There is nothing tying it to their smart devices. It's maybe that they just choose to use it here first or here publicly. Who is to say it isn't used internally and they just haven't said as yet?

In seriousness it is so that Google could update a kernel and any board/platform specific drivers would (in principle) be unaffected, plus all the extra security.

In theory a great idea.

EDIT: It's actually true about the GPL to whoever downvoted that replier, although saying the precise reason for that even remotely publicly would get you shot.

I also want to know why they did not start with seL4 as the base. A Proven micro kernel foundation with real resources to provide all the missing user space pieces would be a huge win for the industry.

They probably started with LK because it had all the drivers from Qualcomm.

https://fuchsia.dev/fuchsia-src/concepts/kernel/zx_and_lk

I wonder the same thing. There's enough there that it could replace the basis for Chrome OS, right? I wonder if that could happen in the near-ish future.

Maybe also it could find its place in Google servers, and more... eventually?

I'm all for supporting competition on the OS space, if I had time I would invest into developing software on one,

however seeing the dire situation on the fuchsia dev team and knowing Google track record on abandoning side projects, kinda makes me skeptical on embarking into it.

The longer this project seems to...drag on? without "replacing android" the more I get a sense it seems like somewhere that Google engineers who aren't "good enough" go to "rest and vest"

As far as I'm aware it runs on exactly one Google product, because they aborted attempts to shoe-horn it into some others?

I would give them the benefit of doubt that there are forces outside of the control of those who are really building these things. Frankly, kind of presumptuous to call them inferior engineers. Working on an OS, particularly one that has to out-perform ios and android isn't simple.

When you see the massive progress being made by say the Redox project (Rust based microkernel), it feels there has to be something internal which is preventing the project from advancing.

What mass market hardware is Redox being shipped on direct from manufacturers? Fuchsia is shipping on the Nest Hub, where is Redox?

Redox is made by volunteers, not a trillion dollar company who decided to install their own OS onto their own hardware.

Point being is that making a micro kernel OS should be fully within their wheelhouse. If Google is not getting it done, there is some process or internal king maker which is preventing it from happening.

But they are getting it done, it's shipping on consumer hardware literally right now!

Yes it is. It's used on Google Hub smart speakers

What mainstream browser has been ported to Redox?

Android is one of the most successful Google products, why would they announce that this experimental whatever it is is replacing it?

It's not going to replace Android, but it could replace the Linux part that lives under Android.

That’s completely different from “replace Android”, isn’t it?

Downvoted because while I don't love Google, I really don't like passive aggression like this, especially when it's wrong.

> ... the more I get a sense it seems like somewhere that Google engineers who aren't "good enough" go to "rest and vest"

The rest and vest thing is rare but the exception is usually people in acquihires they don't want to let go of. Usually they're the opposite of what you're imagining--very smart people they just don't want working at other companies.

The ones who aren't good enough are released. This is 2023.

Reply on news.ycombinator.com