This is a huge development IMHO. If executed correctly, this has the potential to change how we look at the web. This is something that can be the starting point of web 3.0.
[EDIT: I agree. The example applications given are not that appealing.]
I have an opinion that instead of some crappy and difficult to use address like unite://mymac.chrismills.operaunite.com/ they should register a TLD (I guess it is possible. Isn't it?) like .OP or something and start using that. So my address will be simply niyaz.op
I think, (as always) someone else will come along and do it better than Opera does and opera will be forgotten again. Sad, but true.
Anyway I love the Opera for these innovations. They are the pioneers in many cool things in the web. Kudos guys!
This doesn't change how hackers look at the web. There's a reason we use servers -- 24x7 availability with network redundancies, backup power, etc, etc. Most hackers I know can run servers off their local box by using either static IPs from their provider or a proxy service. But many don't because of the reasons above. And now that there are so many great (mostly free) turnkey hosting services like App Engine and Heroku, it's never been simpler to offload all the server issues to a cloud.
The main target audience for this would be users who can't use any of the cloud hosting and also don't need high availability. My folks live by a lake with satellite broadband -- bad latency, frequent dropouts due to weather, and a host of other issues, like simply not wanting to keep their computer on all day, every day. They couldn't use this service. Other users would likely confront at least one of the issues. I've just browsed the Opera material, but I don't see a database/datastore -- maybe submitted data is supposed to go to disk? (The blog example has submitted forms going to memory!)
This is not about reliability. We are not talking about traditional servers and applications.
This is a huge development because when everyone can host applications for free in their own browser we will soon start seeing new paradigms in the way how we use the internet and how applications are served.
I am not smart enough to forecast HOW this will be done, but I have a gut feeling that this will lead us to something really game-changing.
What kinds of apps is this supposed to unleash? If the application is sufficiently complex to require programming or a database, which is true of just about any non-trivial app, then the user is savvy enough to host apps for free in several cloud hosting services with very easy deployment. If we're just talking about file or photo sharing, then there are other services like Dropbox that work well and don't require constant connectivity and compute power on the client side.
I'm typically a fan boy of new tech, but other than a nice proxy service, I don't see why this is revolutionary or game changing. Time will tell, but if there really is a market for something like this, it'll be served by a really cool and simple app development tool which autosyncs to a cloud. Building the app in a brain-dead way is the chokepoint for people coming up with personal apps, not the server infrastructure (at least now in 2009). Why bother with client-side computers for a web-facing personal presence?
OK, I'm willing to take downvotes here, because I'm going to slap HN a little. Why are many people upvoting niyazpk's comments (+14/4 vs other +1 comments before I head to sleep)? To sum up his arguments: it's cool and he's got a gut feeling that it's a web game changer. Meanwhile, there's a bunch of more critical comments that point out Opera's insignificant examples and reasons why the developers that can most use this technology would be better to use more reliable cloud hosting. And these comments get some ups then get pushed back to 1 point. Aren't we supposed to be more analytical here? If you're going to upvote something, make sure there some substance behind the comment rather than what's tantamount to a "that's cool, Opera rox."
Turning a browser into a web server is not a big leap in technology. That is not a hard problem.
Why I am excited about Opera Unite and not about this firefox extension is that Opera themselves thinks that this will change the face of internet applications. They are excited about it and they will persue this to the end. That is the whole point.
It is not about technology. It is about how you push it and persue it. It is all about your vision.
]
Well, they do show how to create a Blog (without any kind of database server setup and believe me - it's very simple!), although I need to find a way to save it between sessions (when you close Opera - entries are gone). They have sandboxed filesystem, so there's probably just a way to dump JSON to a file and reload on start.
Which is cool, but again when you decide to leave for vacation for a week and take your laptop with you to look up dinner options where you're going and don't leave it running all day in your hotel room - no one can read your blog! Am I missing something?
That's also what I'm looking forward to see.
Personally my solution here would be distributed P2P: encrypted content dispatched through neighbours/servers (for some part of the content only, you could choose, for your blog but not necessarily all your shared disk).
I think this + some distributed P2P could well be the web3.0: a new way poeple could use and share their data (a huge open facebook killer).
You are maybe to young to have known it, but HyperCard (http://en.wikipedia.org/wiki/Hypercard) is typically the kind of application that could be distributed via Opera Unite.
It was more like a database that you could program graphically and that was accessible for newbies.
And actually, Ward Cunningham wrote the first wiki with it.
One of the common use-cases that pops into my head is as an equivalent for Flickr except massively de-centralized.
There are obviously many dis-advantages to using a non-centralized system, largest in my mind is the loss of the community feel, but for some people the fact that you are 100% in control of how your information is dealt with might be enough to win them over.
It's likely too that other developers will think of new and exciting things to do that will be unique to the platform, one killer app could make Unite really shine.
That's the biggest feature that pushed flickr to the top of the photo sharing heap. Which is why this is a losing proposition for just about every existing-app-killer idea. Maybe it'll provide a dead simple way to do something new, but any system that relies on consistency and availability (most social networking apps) will suck.
How many 404's before you delete the bookmark and stop coming back?
What kinds of apps is this supposed to unleash? If the application is sufficiently complex to require programming or a database, which is true of just about any non-trivial app, then the user is savvy enough to host apps for free in several cloud hosting services with very easy deployment.
I think you are looking at this the wrong way. Don't think permanent services or sites. Don't think technical people wanting to host solutions.
Rather think end-users just wanting to get stuff done, minimal fuzz.
Anyone can install a local application like MSN, but your mom and dad wouldn't be able to deploy the needed infrastructure on a server. If you can install a application and the required "server"-infrastructure by simply adding a widget to your browser, and invite people to it by simply sending them a link, that is quite a major leap in simplicity and will allow for entirely new forms of spontaneous collaboration.
I'm not saying this is the second coming of Christ and it may even end up as a total dud, but if you are looking at this from a hacker/tech perspective and all you can think of is "but hosting stuff in a datacenter is better" you haven't really gotten what the idea is all about.
will allow for entirely new forms of spontaneous collaboration
... that require pre-packaged apps to be added as plugins to mom and dad's browser, then e-mails sent containing urls to their new service, then mom and dad not turning off their computer and having a good enough connection for whatever type of app they've installed. (Rule out real-time with my parents' place.)
As opposed to our current model where the app writer hosts said new form of spontaneous collaboration (e.g., twitter, skype) on servers and mom and dad discover the service through word of mouth. Then mom and dad install clients where appropriate (e.g., twitter on both computer and cell phone, or skype on macbook with built-in camera). And then there's Google Wave which seems like a better platform for new forms of spontaneous collaboration (more robust, easier to determine recipients/access, open & published protocol).
The main benefit of the Opera system seems to be the app/service store, which adds discoverability, and the url namespace through its proxy service. If there is a market due to these strengths, I'd expect Google's new Wave platform to do better.
Do your folks currently have their own server, how about your cousin with the Nintendo DS (Opera browser) or perhaps you have a server on your iPhone (serving by bluetooth maybe?) ... my post from the 14th, http://alicious.com/2009/opera-about-to-change-the-world/
You forgot file sharing. It's the easiest way that I know that's guaranteed to work for both Internet and Intranet. The rest are really useless, I agree, but the file sharing service just works.
To attract a truly formidable user base, though, (as in, beyond college students and technophiles) I think they need a killer app which has nothing to do with "a file."
Unfortunately, I think that standard computer/web users know or care less and less about what "a file" even is. Is it a photo from their Photo Library, or an album, or the photo library itself? Is it a piece of music they have which shows up in their video editing and music listening programs and also in their personal web space? Where is that file even located?
If Opera's file sharing system also can hook in to everyday users' existing media and document databases, it might go somewhere, but watching the moms, dads, and other casual users of the world have to actually find a file on their computer in this era of iTunes and iMovie and Office 2007+ managing and obfuscating where things are, is fairly painful and distressing. I like that there are systems which let people use computers in this abstracted way which is a lot closer to how people naturally think about things, but I don't like that the abstraction is one layer up from the OS itself, so that when one actually has to do things like, say, find where one's MP3 player is storing your copy of Kenny Loggins' The Danger Zone, for many people it's no longer an instant "a ha" moment.
Oops this post isn't entirely about what it was originally about anymore. Sorry for the ramblin.
I think it might be possible to use CNAME DNS record (athough needs testing) for that task or use modern registrars redirect domain / framed domain features.
update: Just tested, no, CNAMEs don't work for Unite.
I think that file (not only mp3s) sharing is the ultimate goal of the internet. The current state when we have big servers to which clients connect is just a coincidence and is temporary. The future is fully distributed :)
Although I could not agree more, one caveat here - hosting a "server" of nearly any flavor is a violation of most current ISP's TOS. I know we do it and get away with it (and its one of the dumbest rules ever), but they will terminate your service from time to time until you agree to obey - it has happened to me twice over a simple HTML/Javascript web server.
The idea of making a web browser that is also a web server makes a lot more sense than the current trend of cloud computing. The privately owned data center concept is unsustainable, and defeats the concept the internet, which is distributed data.
Eben Moglen, I believe, made some great points about the huge increase in distributed computing power (ie, the power of the laptops in the audience) in this talk:
Data centers get a lot of press, and it is warranted, but regular old desktops and laptops (and smartphones, etc!) are incredibly powerful and underutilized. The internet doesn't have to be cable tv, it can be the telephone game with checksums.
The privately owned data center concept is unsustainable
It's not clear what you're saying here, but I think I strongly disagree. Data centers have real economies of scale (both in dollars and in availability) which shouldn't be ignored. Web 2.0 "freeconomics" may be unsustainable, but you don't have to use that business model; you can make data centers (or "the cloud") work for you.
...and defeats the concept the internet, which is distributed data.
IMO the concept of the Internet is to move packets, and servers are equally legitimate as PCs.
Distribution means not a single point of failure. So both cloud computing and maybe Opera Unite are advances on that.
But saying cloud computing is not the direction I believe it wrong. The idea of cloud computing is really adding more distribution to your network. So you can have cheaper servers all over rather than just your servers at one data center with a remote data store that never has been checked.
Opera Unite is really just the IRC and a slice of P2P. But there will be huge reliability problems as with any distributed P2P system.
Not every piece of data is a file. For instance, is talking to your friends on Facebook a file? I suppose you could view it as a "you talking = writing, friend talking = reading" type thing, but that's stretching it a bit.
There are other examples that have nothing to do with file sharing, like GPS, Google Maps, and pizza delivery.
The difference is that users aren't sharing data with each other, but rather companies are providing services by giving you data.
And a file is just a way to store a database in order to interface with the OS, not the way records themselves are stored. Point is, the Internet is a lot more nuanced than "file sharing".
I think it's very useful for transferring big files (like movie files for a project or from the last trip) to friends. Instead of using megaupload or having them install a software (like DC++, connecting to hubs and so on) to download it. It's like having your own apache server but way much easier to use for normal people.
Technically this is nothing revolutionary. I worked on a project that included a jetty server and SQLite database as part of the download. For the user it was a single button click and updates were automatic. The purpose of the server was to supply services that are difficult with central servers, like watching your page views to see what you were interested in. The local server talked only to your browser and the central server, but it could have easily talked to your friends. The advantage it had over the cloud is that it was using free compute power and it could be an assistant without violating any privacy issues. Of course, generalizing the service raises other issues. You certainly dont want 50 or 100 servers running on your machine.
Meanwhile I'll be praying this client gets supported and extended, otherwise I fear we're in for another eMule. . .
Although I have a lot more faith in the Opera community than preWeb2.0 p2p applications. . .I better stop though, it's only 5am and I'm already speaking in hyperbole.
One of the nicest points of this idea is that it leverages existing technologies really well. eMule requires both parties involved in the transfer use the same software. With Unite the receiver can just use their normal browser.
I think that will go a long way to helping spread the idea. People will ask friends how they're managing to share their content like this and word will spread.
The Opera Unite API might be the other salient feature what everyone here is missing. Their India-evangelist pointed me to it on twitter.
Opera is likely counting on 3rd party developers on making the cool/killer apps. Things like setting up a photo gallery of kids now doesn't require you to sign up for any 3rd party service anymore. This is the anti-cloud in the sense that you can own your data. (pending a security audit of their n-tier architecture)
Disclaimer: I'm not at my desk and haven't tried the new release yet. I may be way off base with respect to the API's capabilities.
Opera is likely counting on 3rd party developers on making the cool/killer apps. Things like setting up a photo gallery of kids now doesn't require you to sign up for any 3rd party service anymore.
Indeed, but I believe Opera could get more mileage out of Unite if it targeted office/corporate workers. Apps would be more profitable for developers, and non-technical people at work could definitely use easier and more versatile collaboration tools. There is potential there.
This would be an extremely hard sell. Getting IT to install a non-supported web browser? Just so random employees can set up random servers on the local network? This sounds like a nightmare to the people who have to support it.
The Opera Unite API might be the other salient feature what everyone here is missing. Opera is likely counting on 3rd party developers on making the cool/killer apps.
If I wanted to build a desktop app that contains a Web server, why would I use Unite over Jetty/Mongrel/etc? The proxy is adding some value, but how much?
I don't like that Opera is trying to fit in everything into the browser. I prefer apps per tasks, don't want a OS in a browser, nor browser with the OS.
Was expecting more of Opera, I don't find this very innovative.
Agreed, I wee the things coming out of MozLabs a lot more innovative - "the future of the internet", if it can be predicted, seems to be along the lines of Moz JetPack, not Opera Unite.
I've been using Opera 10 since the alpha was released, and I can rarely stomach opening up Firefox anymore, although I need to for firebug. Firefox just seems so... bloated. But I digress.
I don't use Widgets, but I do use the BitTorrent client quite a bit. It's a lot more straight forward for smaller downloads than opening up uTorrent. For larger downloads, I do use a dedicated Torrent client, however.
Actually, P2P communication inside any browser has been possible since the release of Adobe Flash Player 10. Two Flash 10 clients can directly communicate with each other over most NATs and firewalls (in contrast to Opera, Flash 10 supports NAT punching).
Just over two weeks ago I've launched a web-based service that uses these capabilities of Flash 10. The service enables users to send files directly between each other. No software to download. If interested, you can check it at http://www.FilesOverMiles.com
I think the main thing is that it just works. You push the button and you get your URL, no matter whether you're behind NAT, router or (probably even) firewall.
It would be interesting to know about such a capability of computers behind the router, because it would imply that the router is not an effective firewall - which I had assumed it is. Although to be fair, I suppose you can usually disable uPnP in the router (which should be the default imo).
I'm just guessing here, but I'd say it is based on a similar scheme as skype - the browser opens an initial route to the opera server, which essentially stays open (think ajax) while the browser is running, and then the server can route the requests back to this link. Any NAT is taken care of in the initial connection, so as long as it isn't broken, any "incoming" request will in fact be a reply to the original open connection.
OK, if they do that, it would work - but then it all would depend on the Opera server. If I create my own applications, I don't want to depend on some Opera service being up and running.
At the very least there would have to be some open source application that I could install on my own server.
If you have a UPnP capable router, it does allow apps behind the firewall to switch on forwarding for ports they need. It's often even enabled by default: There was some exploit recently that used that mechanism.
There are reasons we don't run web servers on our laptops. Putting the server inside the browser instance just makes these reasons more acute.
(Hey guys, I found this awesome site earlier today -- it has 16 hours of downtime a day, but scheduled irregularly, and any given session might be terminated at any time if the owner gets kicked out of the coffee shop where he is getting Wifi or quits out of Opera while playing WoW. Aside from that, yeah, awesome site.)
And they want to write web applications on this? Oh, THAT is going to be fun.
> There are reasons we don't run web servers on our laptops.
Ehrm, I do. It's great for developing on localhost and some quick sharing with someone near me.
> And they want to write web applications on this? Oh, THAT is going to be fun.
I don't think they're aiming to "host" the next Google... From the site I gather that they want to run applications that are for the user of the browser to enjoy by doing something together with his/her buddies.
There is no mention of security in the PR article.
Wondering how easy or difficult it would be for someone to hack into my local opera web server? Possibly the answer is as much as they can hack into your normal web server.
I would guess a bulk of internet users would fear to use this service. They would continue to upload than share local directories online.
There are a whole range of fundamental differences between securing a client application and securing a server application.
I'm not saying that the Opera browser will de facto become insecure, but it's just opened itself up to a potentially huge range of unknown exploits - all servers or software services require some knowledge and configuration to make them secure.
I wish them well, but I foresee this causing a lot of security problems.
"For now, entries are stored in a simple array, so will be lost when the service is restarted, but it wouldn’t be so hard to extend the example to provide a means of retaining the blog entries."
This could be huge combined with webhooks and a JS asynchrounous DOM modification capability. Think about a web-chat app that does asynchronous true push chat. No Comet and heavy JS.
It would work like this : you would connect to the web chat. Your browser would give the server an URL on its internal server where the server can post updates. Just like a webhook. The internal server would run a JS callback each time it got something posted to update the page. BAM! True bidirectional push!
A pity that it would need to be standardized by the W3C to have a wide userbase and thus be a viable option. And standardization would take how long ? 20 years ?
Didn't Windows Home Server do this a long time ago? They also had easy file sharing and gave you your own subdomain with a web interface to get at your content. Although Unite is a lot easier to configure/start and you don't need a dedicated device, I still don't think the core audience is the average user.
It's a little awkward if you need to leave your laptop on 24/7 for your friends to access...For the average user, an internet service is more than enough.
I think that internet based services are so great because they free users from technical details like "files", "sharing", "web servers" etc and let them focus on features - "communicating", "editing". That's why it's strange to see Opera promoting the old all-in-one idiom leaving user with a bunch of files on his personal computer again. Maybe I just don't get all the details and will gladly accept counter-arguments.
I use Opera. Have been since version 4.x, and haven't ever changed. I only keep Firefox around for Firebug and Flash (nice place for Linux Flash to crash in a closeable sandbox).
Firstly if this takes off it will make my job insanely difficult (nigh impossible) because identifying where content came from will probably be even harder (aka impossible).
Secondly it will make my job even busier because it's a hot bed for virus distribution, cracking and intrusion... they better damn well secure this! :D
From my perspective I suspect this will be another Google Hello episode :(
With the direction of broadband ISPs being going to bandwidth caps, 'hosting' on personal PCs is not something I'd be looking at getting into at this point. It's kind of like why I would choose using Vimeo or Youtube to host a video instead of my own web server, bandwidth is not going to be free. I never see this getting beyond a miniscule niche.
Now I have to worry about browser uptime? People don't need a server in their browser, they want reliable services that are easy to use. This could easily have been a nice user interface to a "launchd" and some services. Server side JS might be nice, but there is no benefit to it being coupled to a browser.
It's a different approach, but to the third-party developer (and the end-user) it definitely has some overlap with Google Wave. Interesting to see the two different approaches to web-based collaboration, and it'll be even more interesting to see how they compete and/or combine.
I had this idea in 2002, except searching of this network via the Gnutella network. I called it the Transient Web.
I came to the conclusion that this would just create fractured webpages, and a site one week could be gone the next. This is really bad for things like Archive.org to keep track of.
I doubt anyone will remember this, but way back around the .com bubble ICQ actually had a feature built in where you could host a website off of your computer using it and your contact list could access it. I don't think I really used it, but it was a neat feature.
The EXE installed all the text with its localisation IDs instead of the actual text it is supposed to be localising (eg. D_SERVICE_INSTALL_INTRO). Does this happen for anyone else?
Now when someone asks me what the phrase "jumping the shark" means, I will have an example to give them.
EDIT: A second look at Opera Unite tells me this is a neat idea, but only if it ships as a separate product and not as part of the browser. Imagine a one-click webserver you can use to share music, files and photos from your PC without having to put them "in the cloud". Opera could have charged for this service, and people would have paid gladly.
Putting this in the browser is, IMO, stupidity on a whole new level.
Downloading an external webserver or ftp is not a problem for most geeks. But it takes configuring, knowing what you're doing, etc.
The goal of Opera Unite is rather to bring this possibility, to another audience, to have easily such services.
this is actually really interesting. I can see a lot of new apps coming from this. I just installed it and tried it and so far I like it. The fridge app is a good idea but needs some design work. Photo sharing worked, its not just a file enumeration, they make pretty web galleries and there's an interesting chat app too.
What does this have to do with Opera Unite? It has the word "server" in the API name. Does that mean I can link to Apache and say that they beat Opera to the idea? Opera's not saying they came up with the first local server; they're just giving the end-user the ability to serve their own content easily and without relying on the cloud.
I think implied was the fact that content serving from browser has been available through Gears (which a lot of people have installed), but it hasn't quite caught on. The difference (as I see it) is that Opera makes it REALLY easy (I can start playing with Opera right away, whereas Gears API looks scary at first sight).
Opera Unite and Gears LocalServer are completely different. (I'm using the LocalServer API for the application I'm writing at work.)
LocalServer saves a set of static files (e.g. HTML, JavaScript, CSS, images) from the web. When the browser requests those files later, they will be served from the LocalServer disk cache instead of the remote server. So the pages will load faster and can be available offline (if the pages are designed to run without accessing any other network resources). It's like a browser cache on steroids. For an example application, see the latest version of WordPress, which has an option to load the admin UI into a LocalServer cache.
LocalServer lets you serve requests from disk to browser on a single computer, while Unite lets you serve requests from one computer to another over the internet. LocalServer is useful for offline work, while Unite is useful for communicating on a network.
If you used Unite for a single-user application (where the client and server are all on one machine) then it might provide some of the functionality of Gears LocalServer. But I suspect that the use of a central proxy server in Unite would eliminate both the performance benefits and offline features for this scenario.
The localserver in Gears is to serve offline, local content to your browser for specific websites that are written to use Gears. It doesn't do anything even close to what Unite does.
[EDIT: I agree. The example applications given are not that appealing.]
I have an opinion that instead of some crappy and difficult to use address like unite://mymac.chrismills.operaunite.com/ they should register a TLD (I guess it is possible. Isn't it?) like .OP or something and start using that. So my address will be simply niyaz.op
I think, (as always) someone else will come along and do it better than Opera does and opera will be forgotten again. Sad, but true.
Anyway I love the Opera for these innovations. They are the pioneers in many cool things in the web. Kudos guys!