Google Says It is Expanding Fuchsia's Open Source Model (googleblog.com) 79
New submitter RealNeoMorpheus shares a Google blogpost about Fuchsia -- a new open source operating system that has been in the works for several years: Fuchsia is a long-term project to create a general-purpose, open source operating system, and today we are expanding Fuchsia's open source model to welcome contributions from the public. Fuchsia is designed to prioritize security, updatability, and performance, and is currently under active development by the Fuchsia team. We have been developing Fuchsia in the open, in our git repository for the last four years. You can browse the repository history at fuchsia.googlesource.com to see how Fuchsia has evolved over time. We are laying this foundation from the kernel up to make it easier to create long-lasting, secure products and experiences. Starting today, we are expanding Fuchsia's open source model to make it easier for the public to engage with the project. We have created new public mailing lists for project discussions, added a governance model to clarify how strategic decisions are made, and opened up the issue tracker for public contributors to see what's being worked on. As an open source effort, we welcome high-quality, well-tested contributions from all. There is now a process to become a member to submit patches, or a committer with full write access. In addition, we are also publishing a technical roadmap for Fuchsia to provide better insights for project direction and priorities. Some of the highlights of the roadmap are working on a driver framework for updating the kernel independently of the drivers, improving file systems for performance, and expanding the input pipeline for accessibility.
I don't buy it (Score:5, Insightful)
Fuchsia is a long-term project to create a general-purpose, open source operating system [...] Fuchsia is designed to prioritize security, updatability, and performance...
Sounds like a pretty general OS alright.
So... what's wrong with linux?
Nobody in a sane state of mind writes a general-purpose OS in 2020 just because he thinks the world really needs one. There's always another reason involved.
There's a golden rule: people may give stuff away for free - happens all the time. Buy nobody pays money just to tell you that he has something to give you for free.
So whatever Google has in mind, it's probably not good for for me.
Re:I don't buy it (Score:5, Insightful)
So... what's wrong with linux?
NIH
Re: (Score:1)
Not root even after you bought it (Score:2)
I think grandparent's refers to the habit of certain manufacturers not to give sudo privileges even to a device's owner. This happens on mobile devices and video game consoles, for example. I believe it's ostensibly intended to allow a mobile or home entertainment device to be engineered for reliability and security at the expense of flexibility.
Or maybe device drivers in user space (Score:3)
I have heard in the past that most Android kernel vulnerabilities are in device driver code (Qualcomm being perhaps the largest offender). Were it possible to run these drivers in userspace with reduced privilege, a great many exploits would not be possible. I found an old paper [core.ac.uk] with some evidence for this.
Famous microkernels of the last decades have definite security advantages to monolithic designs, in the ability to reduce privilege and isolate critical components. Minix, QNX, and GNU HURD are likely the
Re: (Score:3)
Re:I don't buy it (Score:4, Insightful)
I suppose GNU/Linux works almost everywhere but it is not great in some places like desktops and laptops.
Linux's weakness on the desktop is due to a lack of familiar applications, not its qualities as an OS.
Re: (Score:3, Insightful)
Even without dealbreakers like that, there is also the fact that nothing remains stable for long, outside of the kernel userspace API, everything else chops and changes so rapidly that it is a support nightmare for developers and te
Re: (Score:3, Insightful)
The fact that Linux still falls over copying files to/from USB drives, hanging the entire UI as everything interactive grinds to a standstill is a pretty glaring issue for a desktop OS to have.
I get that with Windows - assuming it manages not to crash with large files - but I don't see it on Linux.
Even without dealbreakers like that, there is also the fact that nothing remains stable for long,
Again, much more of a problem with Windows.
Also, apps written once, to be left alone to “just work” are a dead end since distros ditch support for much needed libraries without caring what gets broken. It does not make for a good desktop system compared to the competition
Maybe you're using some weird distro. Or are a troll. Basically, you're talking bollocks.
Re: (Score:2)
I ported a very large system of apps from AIX to linux on my desktop Kubuntu 18.04 system as a proof of concept. When the time came to go more official with the port, I brought everything over to a Kubuntu 20.04 system. Leaving aside the odd choice of Kubuntu for a headless server, I was pleasantly surprised that most everthing worked without even recompiling. But, for some reason Ubuntu switched from ncurses 5 to version 6, and it wasn't backwards compatible. I don't remember if I just had to recompile
Re: (Score:2)
The fact that Linux still falls over copying files to/from USB drives [...]
I get that in linux with badly supported hardware. On hardware with good drivers that doesn't happen. It's not a weakness of the kernel itself, but of hardware manufacturers. Fuchsia will, if anything, be worse at it, because now it's going to start off as a 4th OS kernel (besides Windows/Linux/Darwin) to need drivers.
Not every obscure SD card reader manufacturer is going to jump through hoops th write yet another driver.
Even without dealbreakers like that, there is also the fact that nothing remains stable for long, outside of the kernel userspace API [...]
See, but we're talking about the kernel here, because that's where Fuchsia is different
Re: (Score:2)
Not every obscure SD card reader manufacturer is going to jump through hoops th write yet another driver. /.
SD cards do not need drivers. The device you plug them into does.
And no idea why this driver issue always comes up on
I suggest to once write one.
The "hard part" in a driver is to address the actual hardware. It is not the interface to the OS.
If you have written a driver for Windows for an obscure USB device, porting that to Linux or Mac OS can be as short as a day and a recompile. Because: you only ha
Re: (Score:2)
SD cards do not need drivers. The device you plug them into does.
External SD card writers implement the interface of the USB mass storage class. By contrast, it appears some laptops' SD card writer built into a laptop isn't mass storage but instead a lower-level interface to the 4-bit protocol, as might be necessary to access SDIO and the SDMI (DRM) features that used to be SD's selling point over MultiMediaCard (MMC).
And even if a computer's internal SD card writer is a USB mass storage class device, a kernel's support for USB storage might be slow in general.
Re: (Score:2)
The fact that Linux still falls over copying files to/from USB drives, hanging the entire UI as everything interactive grinds to a standstill is a pretty glaring issue for a desktop OS to have.
I've only ever seen that on Windows.
The same issue can be reproduced with heavy I/O to/from an SD card on Android too. Even without dealbreakers like that, there is also the fact that nothing remains stable for long, outside of the kernel userspace API, everything else chops and changes so rapidly that it is a support nightmare for developers and techies alike.
I've only ever had that nightmare on Windows
Also, apps written once, to be left alone to “just work” are a dead end since distros ditch support for much needed libraries without caring what gets broken.
Once again, that is something that I've only ever experienced on Windows.
It does not make for a good desktop system compared to the competition
Seeing as how you've perfectly described the competition, I think you're high.
Re: (Score:2)
Seeing as how you've perfectly described the competition, I think you're high.
I've just used Linux for a long time and know the problems it has, same with Windows and macOS. The competition does not have the problems I've described under normal, default operating conditions, while Linux does.
Do you trust LWN? [lwn.net] The USB/SD file copy crashing problem got introduced into Linux in 2011 and still hasn't been fully resolved in most distros default configurations in 2020, despite multiple attempts (in 2012, another in 2014) to fix it, the problem still rears its ugly head. You can comp
Re: (Score:2)
I suppose GNU/Linux works almost everywhere but it is not great in some places like desktops and laptops.
Eh? It works great on my work laptop, my wife's laptop, and the main desktop machine.
Re: (Score:2, Funny)
So... what's wrong with linux?
System D.
Re: (Score:2)
It's a microkernel architecture - which means it should have better security than a monolithic kernel. It should also run slower, but that's apparently a tradeoff Google's willing to make.
Re: (Score:3, Informative)
microkernel people make that claim about security but never deliver a general purpose operating system that gets accepted for common use, much less prove their claim.
Re: (Score:1)
You mean like macOS - they use a Darwin microkernel.
Re: (Score:3, Informative)
Darwin is an OS. Its kernel is called XNU. It is not quite a microkernel, it's a hybrid.
Re: (Score:2)
and "hybrid" isn't a microkernel at all, the core services are all in kernel space. As Linus T says, "hybrid kernel" is just marketing term. Apple's kernel is monolithic, because microkernels can't cut it, way too much performance drag when lots of messages need passed and context switching happens.
Re: (Score:2)
No, Mac OSX borrows some code from Mach but it is not a microkernel. The services are in kernel space, it's monolithic. Or as Linus says "hybrid kernel" is just laughable marketing term.
Re: (Score:2)
Re: (Score:3)
And really Linux has user space driver support, so if one has complete control over the hardware, hypothetically one could push a lot of the drivers into user space and accomplish much of what a microkernel manages to do. Still, monolithic kernels will have a lot of activity hanging out in ring 0 and ring 1, whereas microkernels shove most of that activity into user land. That doesn't mean that insecure drivers don't exist, but you won't be getting the biggest vulnerability of traditional monolithic kernels
Re: (Score:2)
Micro Kernels are not about security. That is just nonsense. The original idea was that you can do system development better on a micro kernel architecture as everything, that is now a driver and is usually linked into the heavy kernel, is: a micro kernel itself. So if you are working on EXT-2020 filesystem, you simply start the micro kernel for it, and test/debug it on your current OS, without the need to build a full kernel and deploy it on a test machine. However with the rise of VMs and Hypervisors and
Re:I don't buy it (Score:5, Interesting)
Maybe what's wrong with Linux is us. We're so used to it we can't imagine anything being better, and can barely be bothered to change even when Linux itself gets a little better.
I agree that if all you're going to do is make another Unix variant then it's hardly worth the trouble. Linux works well enough to get by with. But maybe there's a way of thinking about the problems an operating system solves for us that's better for what we're doing today than a Unix variant.
I started in Unix back with Version 7, circa 1980, and I remember how liberating the "everything looks like a file" felt. Maybe there's some new metaphor that would make our modern work a little more pleasurable, or makes it easier to do all the things modern Linux *can* do but most of us can't be bothered with.
Re: (Score:2)
This. If you look at the design of Fuchsia it more resembles Android than Unix. For example it uses objects instead of the filesystem for interacting with the kernel. It's mostly event driven but in a way that is more similar to Android's intents than a typical desktop OS message pump, and that extends right down to the lower levels.
Android is built on top of Linux, it's not providing APIs that wrap Linux functions but rather building its own complete OS on top. iOS is somewhat similar, although at the core
Re: (Score:2)
Well, the Plan 9 metaphor is: everything is a filesystem, e.g. your storage of your email program.
Re:I don't buy it (Score:5, Insightful)
So... what's wrong with linux?
I'd venture to say the GPL. They want a proprietary-friendly license and this means something like MIT/BSD/Apache licenses.
Re:I don't buy it (Score:5, Informative)
Yep. [googlesource.com]
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
* Neither the name of Google LLC. nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
Other than the "AS-IS" disclaimer at the end, that's the whole license.
Re: (Score:2)
At least when Google drops support for Pixel 2 after 3 years*, LineageOS programmers can update their devices to later releases using the Linux source code.
If vendors never release device specific code, you're at their mercy. c.f. Apple and Microsoft.
But it's also about vertical integration. Chromium and Google Play don't really care about the underlying OS to the extent that you're coding to a sandboxed runtime - replace Android's Linux kernel with Fuchsia and an app written in Kotlin would seamlessly tran
They want to avoid the GPL (Score:2)
You know they want to get away from the GPL.
Re: (Score:2)
I'd venture to say the GPL.
Yes, that's my guess too :-)
I'm still pondering for a reason why they're not going to choke on it. Writing a mass-market OS from scratch in the 90s, when the only real competition you had was MS-DOS and the only real hardware you needed to support was a i386 must've been kind of fun. But today?
How are they going to keep that up?
Microsoft is (apparently) moving away from their own kernel because it's such an expensive mess.
And Google is building one?!
My best guess is that (1) downstream vendors like Samsung,
Re: (Score:2)
Microsoft is (apparently) moving away from their own kernel because it's such an expensive mess.
It's the other way around, they're supporting the Linux kernel atop the NT kernel, they get the benefit of hardware manufacturers being able to write drivers that don't need to be in the kernel or rebuilt for every kernel upgrade while also getting Linux compatibility by virtualizing those devices.
My best guess is that (1) downstream vendors like Samsung, HTC, Huawei, LG etc. really, really want a proprietary OS; and (2) it's pretty much necessary in order to implement customer surveillance mechanisms without anyone being able to do anything about it. Currently, with the Linux kernel still being open, you can't do (2), because if you put those in the kernel you have to publish them
This doesn't seem plausible, if they want to do that they can just do it in their device drivers that link to the kernel and they don't need to publish them at all. If you're running a proprietary SoC with all pro
Re: I don't buy it (Score:1)
Re: (Score:1)
What's wrong with Linux? How about 98% people find that Linux doesn't meet their needs for an OS, even though it is free.
Google created Fuchsia to test the idea of a downloadable thin client for internet applications, using a capability based micro-kernel for security, ease of maintenance and updating. And also to test Flutter by building the OS and apps with it.
The fact that Google are relinquishing control shows the exact opposite what you claim. They have no further plans for it, but instead of just bur
Re: (Score:2)
Hmmm, where to start? No, seriously. What's wrong with a monolithic operating system developed in 1969?
So whatever Linus T. has in mind, it's probably not good for for me.
Re: I don't buy it (Score:2)
I was going to say "nicely trolled", but that's not even true. Hardly worthba reply but I'm feeling generous, so here we go.
Linux was developed in '91, not in '69. The API may be older, but that's a different pair of shoes. And the monolithic vs microkernel argument was discussed at nauseam by Torvalds and Tannenbaum in the '90s. Sincen then, the best Minix has become is being used in the Intel ME, and Linux essentially everywhere else - so, we pretty much know who won.
You'd at least have a trace of an argu
Re: (Score:2)
Now that's a weighty argument. Well done sir. 'Cause, as we know, there is nothing that separates the two other than design. Sigh. Google QNX, for example. Linux vs a lot of other stuff reminds me of the VHS vs Beta "fight". The best quality doesn't automatically win.
This is another place to start [wikipedia.org]
Project goals? (Score:4, Insightful)
What do they want out of it???
Even the wikipedia article doesn't seem to have an idea of the goals of the project.
Re: (Score:2)
The main advantages it has over current OS'es is that it's a microkernel - with all of the advantages and disadvantages that entails.
Re: (Score:2)
Re: (Score:2)
> I don't know how the Wikipedia editor came to this conclusion
He doesn't know computer science. Fix it.
Re: (Score:2)
Re: (Score:2)
Probably to try to be more secure than their current monolithic offerings. I'd also assume they're looking for "one OS to rule them all" when it comes to their hardware.
Re: Project goals? (Score:1)
Re:Project goals? (Score:5, Interesting)
The original plan was to replace the Linux+Android combo with a microkernel OS+Flutter. Then first canned the Flutter part [slashdot.org], then reclassified Fuchsia as an "experimental OS" [theverge.com], and now they start accepting outside contributions. Which at first sounds great, especially because Fuchsia was only open source in the loosest sense. It used to be developed on an internal Google repo that was inaccessible for outsiders. But it seems to me that this is more about the Fuchsia team seeing the writing on the wall and hoping that the open source community can succeed where they failed. In my opinion, Fuchsia is heading straight to the Google graveyard.
Re: (Score:3)
What do they want out of it???
Maybe also as a way to step out of C ? I'm a C programmer and I write Linux drivers, so I won't spit in the soup, but one day we'll have to have an OS written from scratch in a more modern and secure language, maybe Rust, maybe something not invented yet.
Re: Project goals? (Score:2)
If it is like Dart and many other Google projects itâ(TM)s to generate a buzz and then quietly join the Google graveyard.
The issue is that there are many Google projects than never find a place in the companyâ(TM)s strategy (whatever it really is) and just end up losing steam, before they have been given a chance. In a certain sense it is good to see some of these projects, since the better ones inspire clones or features to be borrowed.
Re: (Score:2)
It is an replacement for Android on phones and tablets but also aims for Desktops.
Lingering questions about Android... (Score:5, Insightful)
1 -- Every commercial install of Android I've ever dealt with was locked, preventing the user from uninstalling crapware. Whose fault is that?
2 -- Most of the Android devices Ive owned had the problem of filling up with updates until they would no longer accept new updates (including huge updates for crapware that I never wanted in the first place) and being mobile/web devices generally, were no longer practical to use. Whose fault is that?
Android was sold to the public as 'open-source' in that it would better serve end-users needs, and it is difficult to argue that goal was ever achieved... or even attempted.
It did data mining for Google and it did whatever the device manufacturers wanted, and the need for OS-level management tools for the end consumer were ignored.
How is the new OS going to be any better than the old one?
Re:Lingering questions about Android... (Score:5, Informative)
No need to install crappy spyware, it's built in at ring 0: https://fuchsia.googlesource.c... [googlesource.com]
Re:Lingering questions about Android... (Score:5, Interesting)
Android was sold to the public as 'open-source'
This.
I bought the G1, on Google's promises that this was all open source, it was gonna be awesome, yada yada.
And then they did an about face and closed off everything they could. Won't be fooled again by that line from Google; I literally paid for it once already.
Re: (Score:2)
> Android was sold to the public as 'open-source' in that it would better serve end-users needs
https://lineageos.org/ [lineageos.org]
Re: Lingering questions about Android... (Score:2)
Try installing that on *your* phone. The one not in that list, with the locked-down bootloader and lack of drivers.
Re: (Score:3)
Re: (Score:2)
.. The whole point of Fuchsia is that the carrier can provide the baseband code, and lock it so it cannot be tampered with, and the phone vendor can supply the correct drivers, and the rest is unlocked...
If you allow the user access to the baseband code, you'd never get the device approved by regulators.
Re: Lingering questions about Android... (Score:2)
Yeah, sorry, but the baseband processor boots from a partition on your internal storage, that is very much writable right now if you got root access. Haven't checked if it does any verification once the main (application processor) bootloader has been replaced by an unocked one.
Re: (Score:2)
1 -- Every commercial install of Android I've ever dealt with was locked, preventing the user from uninstalling crapware. Whose fault is that?
Strange, I only have two Android device which do not have the restriction.
But the answer is easy, no idea why you do not come to it: the device vendor is at fault, as he locked it down.
2 -- Most of the Android devices Ive owned had the problem of filling up with updates until they would no longer accept new updates (including huge updates for crapware that I never wan
Re: Lingering questions about Android... (Score:2)
A true open source licens does not allow locking down.
This locking down is apparently legal.
QED.
Re: (Score:2)
Yes, and the devices in question have an option to "switch off" lock downs. So it is his fault not to witch it off, oops.
Year of the Desktop! (Score:2)
Re: (Score:2)
I hope it's not pronounced "fucks ya."
Re: Year of the Desktop! (Score:2)
It is.
Vendor lock is the obvious goal. (Score:2)
Build a de-facto walled garden and you win. The Apple example is obvious and fantastically profitable.
Re: (Score:2)
Need guide to the interesting parts (Score:2)
Okay, I would be the first to raise a contempting eyebrow at Google's idea of open source and long term viability. However it might be interesting to learn from what they are doing. They have a bunch of smart people and may have interesting takes on an OS.
As an example my 2 minute skim of a random page in Fuchsia's Concept section, Dot Dot Considered Harmful
https://fuchsia.dev/fuchsia-sr... [fuchsia.dev]
1. Top of the page says "Google is committed to advancing racial equity for Black communities. See how." Okay... didn't
Sounds like a Duke Nukem Forever. (Score:2)
Changing goalposts, years without results, big corporate moneybags feeding it anyway... Yeah, it's gonna be a squib.