Web OSes: There Are Really Or Just a Fully-Featured Web Site

[Nathan]

Hello,
These days are very common to hear people talking about the cloud computing and the Web OSes, as example the Eye OS, but in my head I still have this question/confusion, they are considered OSes, or just fully-featured web pages? Because to use them, you still need to have a OS installed on your HD(or running live on CDs or even floppys - as MenuetOS) and a web browser, but the Chrome OS seems to be something mysterious, because it's a Linux kernel running a web browser and window manager.

Best Regards,
Nathan Paulino Campos

[NickJohnson]

The point is that most of the system's functionality (the applications) is online. For both EyeOS and Chrome OS (iirc), the only thing running directly on the computer is the browser and supporting structure like the GUI, network driver, etc. So, it's a little more than a fully-featured website, because it is both a website and a browser, and when you make the interface work like it does in a Web OS, it's harder to call it a normal website. Although I would consider them OSes, thinking they're a good idea is a whole different thing: having Google Docs entirely online is reasonable; having something like gvim or QEMU or Photoshop or TF2 is not.

[Creature]

having Google Docs entirely online is reasonable; having something like gvim or QEMU or Photoshop or TF2 is not.

I wholeheartedly agree. I'm ok with webbased "applications" for small things, but I think moving giant software like Photoshop and such to internet is just not what it's supposed to be. I can't exactly explain why, but e.g. playing Quake 3 where you only get the graphics, but most of the game runs on the server (à la thin client) just isn't right.

[Solar]

The term "WebOS" is a very bad joke.

You can have your home directory / "My Files" online. You can have most, perhaps even all your applications online.

But the OS - bootloader, scheduler, device drivers, IPC, memory manager, event handler, TCP/IP stack (*) - have to live on your local machine.

Borders become watered down when you're doing network booting (because you don't have the files of the OS on your local machine either), but still, you need all of the above code running on your local CPU in order to even access a website.

Even if the only thing you ever run locally is a web browser, even if all you ever see is this website, the website is not running your computer.

It's still the OS under the hood that will decide how efficiently your local resources are used, and how stable your browser will run.


(*): I remember a time when, not only the webbrowser was an application to-be-installed, but the TCP/IP stack was, too. Gosh, show me the way to my tomb, I'll start to rot away any minute now...

[Nathan]

The term "WebOS" is a very bad joke.

You can have your home directory / "My Files" online. You can have most, perhaps even all your applications online.

But the OS - bootloader, scheduler, device drivers, IPC, memory manager, event handler, TCP/IP stack (*) - have to live on your local machine.

Borders become watered down when you're doing network booting (because you don't have the files of the OS on your local machine either), but still, you need all of the above code running on your local CPU in order to even access a website.

Even if the only thing you ever run locally is a web browser, even if all you ever see is this website, the website is not running your computer.

It's still the OS under the hood that will decide how efficiently your local resources are used, and how stable your browser will run.


(*): I remember a time when, not only the webbrowser was an application to-be-installed, but the TCP/IP stack was, too. Gosh, show me the way to my tomb, I'll start to rot away any minute now...

This is why I didn't agree with the term.

having Google Docs entirely online is reasonable; having something like gvim or QEMU or Photoshop or TF2 is not.

I wholeheartedly agree. I'm ok with webbased "applications" for small things, but I think moving giant software like Photoshop and such to internet is just not what it's supposed to be. I can't exactly explain why, but e.g. playing Quake 3 where you only get the graphics, but most of the game runs on the server (à la thin client) just isn't right.

Small applications is ok, because you need to use them online when you don't have it installed on your computer, and need something fast, as I do many times with image converters, use them online.

[NickJohnson]

(*): I remember a time when, not only the webbrowser was an application to-be-installed, but the TCP/IP stack was, too. Gosh, show me the way to my tomb, I'll start to rot away any minute now...

That's technically still true with most microkernels: they may have network stacks in the base system, but they can be removed with only network-related effects. When I get to a beta release, I'll make a version just for you that installs its network stack from a floppy.

Back on topic though: There are other problems with having everything online: the one that I think of first is security. What happens if the main application server is hacked and a virus effectively inserted into every computer currently running simultaneously? Another is energy: if computers take up a reasonable amount of energy locally, what happens when every disk access has to be sent back and forth across hundreds of miles of internet cables?

[Coty]

WebOS rings in my head as a cool idea for a netbook as a second PC... But I would
like a partition so I could do other things. that is why I have come to love the idea
of Splashtop - A small browser OS that lets you switch to your fully functional OS
at any time, I do also like cloud computing but ONLY for backing up my data nothin
else

I think splashtop is better than what google has

[JackScott]

There are two different sorts of computer users: consumers and producers. Producers produce stuff for consumers to consume.

Consumers are probably quite happy with a single-tasking WebOS (I mean, how many people buy iPhones?). All they want to do is read the web and listen and watch. Even we're all consumers sometimes.

Producers are another breed entirely however. They want power (for compiling and rendering), they want flexibility, they want multi-monitor setups, they want 6 hard drives. Most of all they want the ability to run things locally on their own machine. When we're developing an operating system, we're producers. Producers (going with the Apple theme) will be happy with nothing less than a full OS X.

Consumers buy products based on recommendations from the tech savvy people they know. The tech savvy people... are producers. And producers like full operating systems. It's just easier for a technician to fix something they already know about. So they'll recommend the full operating system, and the WebOS will be just a fad. Especially since practically nobody is a full consumer all of the time.

This probably isn't a really well thought out theory... but it's my $0.02.

[NickJohnson]

I can find a hole in your theory. I agree that there is definitely a split in the populations that use computers, pretty much to the effect of "producers" and "consumers", but the producers don't have that much impact on the consumers.

First off, there is a third group of people, commonly known as "marketing", that tries to sell as much stuff as possible to consumers regardless of what exactly the stuff is. Commercials probably have ten times more of an effect on the market than recommendations by tech-savvy friends of people in the market, and the producers of the stuff sold to consumers have little control over it.

Second, as a contradictory example: a reasonable (minority, but visible) fraction of the tech-savvy population uses Linux, particularly those who "produce"; few people who use Linux are not tech-savvy, and few people who are not tech-savvy use Linux; if influence came from producers, a much larger fraction of consumers would use Linux.

[Benk]

The idea is simply to try to get a more secure environment sandbox by limiting the language the OS can run , most new OS use this Web OS , Android ( Java) , Singularity (CIL) . The Web OS are using Java script as their local language. IMHO they are flawed and should opt for an IR representation ( either LVVM IR or something like CIL) and compile on install.

That way like Java script they can provide a secure more environment but unlike them they allow compile on install ( giving you machine optomized code like Gentoo ) and allow a wider range of languages while eliminating risky code. ie RIA like Flash and Silverlight can be supported in a safe manner which a webOS cannot. RIA apps also handle sometimes disconnected scenarios better.

They also use Web keys / capabilities more which is more like an OS the Coyotos but only provide this for the sand box ( mainly because writing a new OS is too much work) .

Note however the old OS per process protection has issues with browsers , if an addon crashses your browser is toast which is just as bad to me as an OS crashing so what do we get from all our fancy OS theory ?

IE 8 has tried per process copies of a browser at a great cost in terms of memory usage but still has the addin issue.

[Benk]

I can find a hole in your theory. I agree that there is definitely a split in the populations that use computers, pretty much to the effect of "producers" and "consumers", but the producers don't have that much impact on the consumers.

First off, there is a third group of people, commonly known as "marketing", that tries to sell as much stuff as possible to consumers regardless of what exactly the stuff is. Commercials probably have ten times more of an effect on the market than recommendations by tech-savvy friends of people in the market, and the producers of the stuff sold to consumers have little control over it.

This works for Apple but MS success has not come from Marketing , their success was basically built on DOS , Windows 3.1 and Excel/Word for windows and there was very little advertising when those products were released ( and X shot itself in the foot by requiring 20K workstations to run on though in hind sight their perf predictions were correct , macs at the time had a similar issue being far more expensive , i seem to remember 5-6K or 11K for a Lisa) . It has come from first mover advantage and being the only product in the user market place and because of that the all important application compatability. Many Linux people still need/have a windows partition .

Their products have also been good enough and to market at the right time for the market to accept it ( eg Windows 3.0 , NT , Word , Excel etc) . Linux rellies on this as well being in the University market with a lot of 70' and 80's unix apps ( grep/find /sed /emacs / vi etc) .

I know many researchers that built OS but gained no traction as it was not 100% windows compatible bugs and all .

Second, as a contradictory example: a reasonable (minority, but visible) fraction of the tech-savvy population uses Linux, particularly those who "produce"; few people who use Linux are not tech-savvy, and few people who are not tech-savvy use Linux; if influence came from producers, a much larger fraction of consumers would use Linux.

Agree , i would add Linux is MUCH easier for admins as it is less feature rich.

[Brynet-Inc]

The funny thing about OS's that try to remove the "native" aspect of an OS onto a website inside a web browser.. is they're all starting to offer "native" SDK's, as an example, both Android (Java) and WebOS (Javascript/HTML, AJAX) support running code to access features of the processor directly for "speed".

The whole "Web Operating System" and "Cloud Computing" trend is a farce, you will always need low level code and a traditional operating system.. and limiting exposure to the OS through "safe" high level interpreters is just going to annoy everyone, and add to the ridiculous complexity of it all.. security through layers of cheese.

So in conclusion, a WebOS is a fancy website.. with no practical purpose, except some fancy buzzwords.

[Colonel Kernel]

The whole "Web Operating System" and "Cloud Computing" trend is a farce

Don't conflate those two things. They are very, very different. I agree about Web OSes, but cloud computing is another beast entirely. Cloud computing has existed for years and already does useful things, unlike Web OSes (see: Google, Bing, Amazon EC2, Windows Azure, Mobile Me, Windows Live, or any other large-scale internet service that handles millions of users, or any platform to build such services).

[inx]

Eh, history repeats itself in all fields and arenas. The computing field is extremely young, so patterns are still emerging. In this particular case, I'm seeing thin terminals.
We started out with single-tasking, single-user systems, and made them multi-user adding thin terminals. Over time, the thin terminals became capable of more as the
underlying system gained capability, until it became cheap enough to have a personal system that was capable of handling what the average user would generally do on
their terminal. The things PCs couldn't do stayed on the mainframes, and you used your PC as a thin terminal.

This, also, evolved (cue the internet). Now, because the networking infrastructure has advanced, we're moving back to thin terminals (or, you could say, thick terminals,
as they run some apps locally. Just like a minivax X terminal would often run simple apps locally, or parts of apps). Eventually, the field will push against the boundaries
of what can be run on "the cloud" using these devices as their terminals, causing an increasing amount of apps to run more locally, while still integrated with "the cloud"
("Native APIs", anyone?). Eventually, everything will be pushed to the device, with only communication and synchronization to "the cloud," leaving us with a new era of PCs.

The cloud is nothing new, it's just an extension and resurrection of an older paradigm. As it has died before, so it will die again. Constructive things will flow out of it,
and any detriment it caused will be healed by time and adaptation. Nothing to see here, move along.

[NelsonC]

The problem with web apps is that Html, Css, and especially Javascript are terribly inefficient for doing that type of thing. I am a million, trillion times more productive using Silverlight than I am using HTML5, CSS3, and Javascript. RIA is simply a breeze when you don't sacrifice performance for the sake of ubiquity.