Slashdot Mirror
Google Releases Source To Chromium OS
Kelson writes "Google has released the source to what will eventually become Chrome OS, and will begin developing it as an open source project like Chromium. The OS differs from the usual computing model by (1) making all apps web apps (2) sandboxing everything and (3) removing anything unnecessary, to focus on speed." Reader Barence adds "Google said consumers won't be able to download the operating system — it will only be available on hardware that meets Google's specifications. Hard disks are banned, for instance, while Google said it will also specify factors such as screen sizes and display resolutions. Google said it plans to officially launch Chrome OS by the end of next year."
So what, does my computer boot up to magic, or are they building a BIOS or LiveCD specific to Chrome?
SJW: Someone who has run out of real oppression, and has to fake it.
If it's open source and available... couldn't interested parties compile it themselves. How are they going to ensure that it's not available for actual use?
This is the infamous network appliance made real. The OS is a simplified kernel with a specific set of supported hardware with a simple interface and no on-system storage for data. All apps and settings are "in the cloud" i.e., on google's servers.
For likely 90% of home users, this will be perfect. A relatively dumb device that only runs a web browser to use web apps (googles or anyone else's provided their signed by google) to do their work.
It takes user-friendly to an extreme and makes everything just part of the web browser experience.
The root OS partition is read only and the selection of hardware is prescribed by google. You can download the source to hack it, but you can't make an installable image as you can't cryptographically sign it for their okay. They're only planning this to be a bought with hardware purchase.
Sound familiar? It should, it's basically the Apple experience made into a net appliance.
... in order for you to use the software when you aren't plugged into the net, you'll have to download some bloated piece of kludge like Gears, which won't actually give you the seamless experience you crave, but will make you wish for the days of Windows 3.1.
The hypothetical "user" will never have a chance to download Google's OS and find it doesn't work well. Google has stated their intention of only providing it with approved hardware.
Now, because they have also announced that it will be OSS(except, presumably, a blob of trademarked logos and stuff), there will most likely be third party builds available; but the sort of people who download third party builds of OSS code can either RTFM beforehand to make sure that their stuff is supported, or deal with it like adults when their unsupported hardware turns out to be problematic.
I plan to move our company to a "dumb terminal" model over the next couple of years. You say that the cost of hardware just gets "shifted", but this is not entirely accurate. I have roughly 60 users. Each machine must be spec'd to handle the biggest workload, even if that only gets hit during some small fraction of the day. For 99%+ of the day, I have a powerful machine doing very little. With a centralized model, I can smooth that out.
But that isn't the biggest reason I am going to this model. I have folks who can be working in our central office, satellite office, on the road, or at home. I need ways to give my workforce the flexibility they need to work anywhere.
From a cost standpoint, PCs are awful. Maintenance is generally more than the hardware costs. Software installation and configuration alone costs us about 1/4 of a FTE. By centralizing, I am expecting that number to drop by 2/3.
Now, granted, my network is either local, or connected by dedicated T-1's except for our road folks. So, while I think this is a great idea for my workplace, I don't think it makes a lot of sense for me at home.
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
Yeah and all webapps which everyone hated when the iPhone did it but this is Google so be prepared to suddenly have it become brilliant and the wave of the future. Hurrah for hypocrasy.
If all else fails, immortality can always be assured by spectacular error.
If you use x86, you've been running all interpreted code since the mid 90s - all x86 processors since the Pentium Pro are RISC processors with an on-chip virtual machine for the x86 instructions. This objection to interpreted code seems to be based on, well, nothing - why should we care what implementation strategy our software happens to be using?
If the OS can't be downloaded, it's attached to the hardware 1-to-1.
The hardware can't cost a penny more than a netbook ($250-300) or we'd just get a netbook.
Removing the harddrive, or putting a small 4gb SD drive, will put it around $200.
$200: Meh.
$150: I'd rush the doors like a Walmart on Black Friday.
After finding this link: http://www.chromium.org/chromium-os/chromiumos-design-docs/security-overview
I'm impressed. I wasn't expecting that much in the way of security in this offering, but I'm actually pleasantly surprised by how much thought Google has put into this, both from remote attacks and local (stolen computer/device).
Three notable things:
I like is the fact that items that log on and use Google's authentication mechanism work online, and offline by using a local cached hash table.
The segmenting of the Web browser. This is something every Web browser should do, so one buggy plugin doesn't mean a completely rooted system.
Very well thought out boot path with initial key values stored in an unalterable chip. Next to a TPM boot, this is a good way to protect against corrupted boot attacks.
My only wish is that the device didn't use an Owner/user priv model. This is just fine for devices and home computers, but when you get to the enterprise where you have to have machines have a "master-root" user (usually an Active Directory) admin, there will be issues.
But only Google's cloud.
Say what you will about Windows, but I can install Chrome, Gears, and bam -- I can use Google's 'cloud' infrastructure.
ChromeOS? I can only use Google.
I'll stick with Windows for now.
On a related note, this is one of the most underwhelming releases I've ever seen. Way to blow the hype.
The price is always right if someone else is paying.
I think I had a knoppix which did exactly : "boot and launch firefox". I don't see the point of developping an full OS when configuring a linux distributin might be enough.
While I extensively use Google's products, I find that GMail is still wanting in terms of searching for email.
Here's why: You search for all mails containing some word...Gmail returns all mails having such a word with no obvious categorization. It would be better if it can return emails categorized as follows:
Those with attachments and what type of attachment it is, those sent last week, last month, last year, 2 years ago etc...those sent by who...and so on.
Right now, the interface sucks big time. Anyone agree? Yahoo does a better job at this.
Uh, anyone that has a netbook that they use for absolutely nothing but web browsing? Which, if everyone else is as disappointed with their netbooks as I am, should be a lot of people. Basically, if you have a computer that you do nothing but web browsing on - Chrome OS is for you.
Will it have Java? I'm wondering because it's still the only way to do decent architecture-independent games without 100% CPU usage 100% of the time.
Or does it support the Native client stuff?
Really, what does an OS need to do? It needs to manage the network, talk to devices and launch applications. That's it, isn't it? By specifying "no hard disk" Google is cutting out a major part of the device chat. Displaying a folder hierarchy is essentially a search, format and display application. They're good at that.
A large part of the Windows code is managing a large variety of devices, from displays to USB devices. If Google specifies the display format, then there's another large chunk of code dropped. The UI is an application, pointing devices are - devices.
Add an IP stack for the network and stick a security layer in somewhere, if you still need it.
By limiting configuration choices to those that have a broad appeal a *huge* amount of OS can simply go away. You have less local IO, less device chat, and no local disk latency to worry about.
People know how long their network takes to react, and will accommodate that. In contrast, a very thin OS will be very quick and will compare very favourably to a thick OS in response. And if most of the IO is server-side in the cloud, you won't see a lot of IO delays (source of most hangs) and response should be smoother overall, because servers tend to have the best IO controllers and enough spindles to stripe (not that Google would resort to actual hard drives!) Where's the beef?
Do not mock my vision of impractical footwear
Exactly.
And that sounds great if you're a programmer right out of college, but Win32 tie-in, specifically with MS Office is still a huge factor in the real world. A big problem here is that the hardware they're targeting will be able to run fullbore Windows 7 just fine.
We've had web-based word processors for fifteen years but Google's web-based word processor is different because it's from Google?
Business. Numbers. Money. People. Computer World.
I watched the webcast and they (Google dudes) were actually encouraging forking and
gave some Chrome fork as an example.
Am I reading this incorrectly? I thought the summary said it *treated* everything as a web app, not that everything was itself a web app. By this I assume that means most of the new apps that will ship with this are written in Gears, and will exist in a sandbox. That doesn't mean every app is Gmail.
Will it be possible to load standard .deb packages of other Linux apps? Probably not anything that depends on Gnome libs and KDE libs itself, but pure GTK apps might work.
http://blindscribblings.com - Tasty pop-culture in conceptual fashion.
Web-based word processors have failed to rise above. So have many other kinds of web-based apps that mimic commonly used standalone apps like spreadsheets, presentation software, development studios - IDEs, paint programs, etc. etc. Most web versions of applications are a poor man's user experience. I know people can come up with exceptions but these are not the norm and likely won't be for a long time. Microsoft is certainly not going out of business because of web-based apps anytime in this dimension.
I think most people will stick with Windows and proper GNU/Linux netbooks
Yeah, this sounds to me like just another try at the failed Internet appliance idea. Didn't work then, and I doubt it'll work now. With a netbook you should be able to run everything Chrome has (as long as you have a browser and a network connection), plus a huge variety of other stuff. For example, on planes I carry a netbook with a few movies and a lot of music; will I be able to use a Chrome device for that?
Maybe if the price were significantly better, I might consider one of those things, but again I don't see how. Netbooks are cheap enough as is, and I don't believe manufacturers will be able to save much on Chrome OS devices.
Price. No HDD and the ability to streamline components because it only does certain things cuts cost, which in turn cuts price. I can see people buying a "Web Tablet" that they already know all they intend to do with it is just what ChromeOS does, and then they have a normal computer for the real work. It's actually closer to the original "idea" of a netbook. A simple, inexpensive booklet that just does the internet.
Here's a cookie... *psst* it's MAGIC
Because I like not having to configure each mail client on each computer I use, having my entire mail history available to me through a browser, and not having to worry about backing my mail up. The notion of a local mail client is quaint.
Fine by me, it's a well known fact that 0.001% of the population holds over 90% of the wealth.
Here's something that should be a real concern for geeks. Right now, the ordinary desktop users who don't really need a powerful computer are buying computers. This means that due to economies of scale, the cost of computers is relatively cheap. Imagine what will happen to the price of "powerful desktop-ish machines with full-featured OSes" if 90% of the computing market suddenly starts using these toys. Start preparing to go back to the days of $15,000 computers. Just saying. :-)
Check out my sci-fi/humor trilogy at PatriotsBooks.
I imagine you're aware of this, but with Linux, you could do all this for free. Citrix'll cost you a ton for client access licenses, and Linux has this kind of thin-client support built in.
Of course, you must be replacing a traditional Windows desktop-centric network, and I guess you have some need for Windows-only apps with no viable Linux equivalent. But don't you wish you didn't have that requirement? Maybe one of these days...
Interestingly, maybe ChromeOS will support a citrix client (or X / NX server). Maybe the new devices built around ChromeOS will make really nice thin-client terminals for more than web-only use. Not clear yet, but that could be enough of a niche to keep the hardware manufacturers onboard. Maybe even Wyse...
Posted from my Android phone. Oh, I can change this? There, that's better...