Slashdot Mirror
UrlHosted Experiment: Host Content Within the URL
New submitter graphicore writes to point out an experimental "unhosted" app that challenges the concept of the URL. By putting the post data after the # mark, the URL is (mis-)used as the data storage. You can store your data within your bookmarks list, host it via a URL-shortener(!) like here: http://goo.gl/DYxr5m or attach it directly to a tweet
I also attached the full-url to this slashdot post :-) This raises the question about who is hosting the content and it will probably break the internet.
This is a quote from Google's shortener policy: "Please remember that goo.gl directs you to content that is already in existence on the internet. This is not content hosted by Google." It could also become a storage strategy for any other web app. The app is GPL v3, no strings attached.
And there's always DNS, too.
... it will probably break the internet...
Oh no Mr. Bill. The Internet will be broken.
.
Give me a friggin' break. Get real.
What's a # sign for at the end of some URLs? I've always wondered that since not all have them! Thanks for the answer.
You still need to point to a base URL that knows how to unwrap the URL hosted content...
So who who host those, knowing that any URL directed there might be mistakenly attributed to content they are hosting? You could make it appear as if such a site is saying ANYTHING... it's like you pre-hacked yourself.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
Now he just needs to get the javascript powering this to fit in a data:// uri and it can be entirely hosted in the url.
How is this different from the Data URI Scheme?
https://en.wikipedia.org/wiki/Data_URI_scheme
All this stuff about non-hosted content, and the image tag points to a wikimedia picture of a kitten instead of a data: URI?
If I have been able to see further than others, it is because I bought a pair of binoculars.
I tried it out.
i think this is kind of cool. it's clear that there's some sort of server thing that interprets the URL and spits back friendly HTMLs. it would be cool if this could be done locally, so alls you would need is a shortened URL and you would get a page of content. it would work well for wikipedia.
How many of these URLs already exist and how much malware are they hosting?
Something like :
data:text/html,<html><body>Hello</body></html>
Will not be "shortended" by a url shortener like bit.ly, whereas the "#" embedding technique will (but then you need to know how to decode it)
"There is more worth loving than we have strength to love." - Brian Jay Stanley
My, what an exciting new way to fuck shit up and break all sorts of standards!
Just cruising through this digital world at 33 1/3 rpm...
and that's all. Why should I allow some unknown host to execute javascript?
Cute concept.
Reminds me of gag websites where you'd put some text in the url and it would generate a fake news article condemning $name_in_url.
That, but more.
Didn't CGI teach us that this is stupid?
Seems IBM WebSphere did something like that. Their default URL's were often longer than a Giraffe's intestines.
http://stackoverflow.com/quest...
Table-ized A.I.
First time I read the summary I was like WTF is this, an advertisement?
Second time, I'm still wondering WTF. Someone needs to go back and read what a URL (ahem, URI, excuse me) actually is. It's a LINK: connecting things... not storing app preferences and shit. That's like, real, old school... before sessions, and cookies!
And that gets to the front page now?
...if the kitten image data was stored in the URL also.
This is against everything we know about cross site scripting. It is like having ?errormessage=text at the end of a URL. We know the security implications of this, and we know not to do it. The potential for abuse is way too high.
Sig: I stole this sig.
I might be subjective as I'm the author of it, but this somewhat remind me of my http://mdoc.su/ project, which is what I call a deterministic URL shortener, or, perhaps, better yet, a semantic URL provider.
The whole source code is an nginc.conf configuration file, and is just a bunch of regular expressions and `rewrite` and `location` rules, available under an BSD/ISC licence, of course -- that's the one that comes with "no strings attached", BTW!
http://mdoc.su/
http://mdoc.su/FreeBSD-10.2/fs
http://mdoc.su/f102/resolvconf
http://nginx.conf.mdoc.su/mdoc...
https://github.com/cnst/mdoc.s...
Google "bookmarklet"
So after about 1 day all the URL shorteners will only store the first 8 characters or so after the # in a URL, ensuring that most actual links to articles scroll to the correct location, but stupid tricks like this don't work.
Get creative, like this.
They could use an insanely high max-age header or the HTML5 Application Cache to make this truly server-less. Each browser would contact the server only once.
I love this idea! Also, I have to add something: If the URL-shortner uses 302 to redirect to the full URL, the content part behind the anchor/hash will not be sent to the server. It will still available to the JavaScript returned from the server that is run in the browser. So the server will not know about the content. (This feature is already used by e.g. OAuth2) In addition, instead of accessing a remote server, the URL could point to localhost, and the user could run the "content unpacking" webserver locally (and maybe automatically prevent any unwanted cross-site requests, since this is the default behaviour of the browser)
Image search thumbnails are all in the url.
A person could use this app to run a blog of sorts, and as popular as it became the blogger would be hosting it on the cheap. You host the app and tweet the shortened URL's. The content is hosted, but not by you. The URL shortener hosts the content. But unlike LiveJournal or Wordpress.com, the URL shortener never agreed to hosting your content. You've essentially repurposed its functionality and subverted its intent.
I'm guessing the various URL shorteners will respond to this very quickly. The hack will end up being as short-lived as it is cool.
quiquid id est, timeo puellas et oscula dantes.
Jesus fa... what the fuck did I just read???
It reads like it's being said by an eight year old girl who's just been given two double espressos and a new kitten.
Dear Mr Fister,
You pretty much stole this from http://hashb.in/ and tweeted it for recognition.
Now you will be known as the smarmy jerk that got Javashitheads excited about breaking conventions.
Further, you just made it so url shorteners are going to get stringent and check if url parameters are obvious encoded (base64, etc).
As others have said, this will be short lived.
I really hope you enjoy these 5 minutes of fame. I will be sure to never hire anyone like you.
BTW, great last name. you make terrible typefaces.
Overreact much?
This will break the T-shirt business! All those whose living depends on selling t-shirts with DeCSS source code, "09 F9" AACS key, etc. as all we need now is a shortened url (white t-shirt + that marker that you have still have from freeing CDs from Sony's key2audio protection).
When the copyright term is "forever minus a day", live every day like it's the last.
There are some examples of URIs that contain their data using the "data" URI scheme.
Since getting at game saves is not something the average user can easily do in most cases on mobile platforms, would this be a useful method for sharing save games?
Just drop a URL to the desktop and now anyone can have full hearts, the champion sword, and the unobtainium underpants.
This raises the question about who is hosting the content and it will probably break the internet.
No, absolutely not. No on both those assertions. In fact, it really clears up who is responsible for the content of the link. As the same host contains both the "link" and the data. People have been converting data to text and embedding it directly into HTML pretty much since HTML has existed. It is neat if often the wrong way to go about it, but also very useful for userscript developers.
Troll is not a replacement for I disagree.
No.
As i felt that is the ISP related issue. if you don't getting internet properly. So try some other internet connection. Host it via a URL-shortener is the some old way technique to saturate the internet connection and redirect according to Google's shortener policy.
That someone on friggin' slashdot has no clue what the # in a url is, and thinks that asking it is easier than just friggin' googling it
# is punctuation, and general-purpose web search engines have historically choked on queries not for letters or digits.
You should write a client application.
If this client application is developed for Windows, good luck running it on a Mac. Or if this client application is developed for OS X, good luck running it on a Lenovo. Not everybody has the money to maintain 14 different client applications, one for each platform, even if they do share some of the code.
Client side input validators LOL.
Where do you work?
Presumably somewhere that realizes the value of validating input once quickly on the client and again securely on the server.
If you're so dead-set against JavaScript, would you rather have to reload all comments to a Slashdot article every time you expand or collapse a subtree?
Three million downloads across how many different platforms' app stores? How do you normally reply when someone asks about wanting to use your client application on a platform for which your client application is not currently available?