If only the http:// part was to disappear, that'd be a huge win.
For example, emails and http:// addresses are already recognized as such in places like the gnome terminal (which is great then for irssi IRC) or even in this forum comments. But can't just domain.com/that/ be enough info to ne recognized as an url?
FF3 works great (for me) with these on Mac, Ubuntu, Windows, etc. What platform are you using?
Based on your error message, I bet it's not FF3 but whatever showed you the link. If I had to guess, you clicked on the link/page through a feed reader or something?
Can you confirm whether it happens when you click directly from this page/site on Hacker News? (which seems to be handling them great)
I tried going from this page directly, and get the same error message (that too seemingly immediately - doesn't even seem like it tries a DNS resolution, though I'm not sure): "Firefox can't find the server at www.%e2%9e%a1.ws."
I'm using Firefox 3.0.5 on Windows XP.
FYI, the other link you gave in http://news.ycombinator.com/item?id=498246 too doesn't work for me: "Firefox can't find the server at www.%e0%b2%95%e0%b2%b0%e0%b3%8d%e0%b2%a8%e0%b2%be%e0%b2%9f%e0%b2%95.com."
Sorry for the accidental downvote-- I have a reflex of downvoting people who reply to the wrong comment and I forgot this was my own thread. Sorry. :(
Twitter is doing a character count and doesn't care about bytes-- if it did, it means that tweeting in non-ASCII languages only gets half the characters. That's not exactly fair.
If people like the service, we'll be adding metrics and even allow people to search for tinyarrows that were associated with a particular domain.
It's more of something fun-- we're not trying to change the world or anything. We make a lot of sites/toys and this is just one of the fun ones to share with people.
I'd heard otherwise about the byte count, but I don't have any authoritative knowledge on the matter
We make a lot of sites/toys and this is just one of the fun ones to share with people.
Awesome. There should be more tiny projects. I recall comments on http://tweetbysnailmail.com saying "this startup will fail". They obviously miss the point :)
The fun part aside, are there any real use cases for tweetbymail?
At the peak of .Com days, there was a startup that promised to bridge e-mail and snail mail: one can use their service to email people who don't have an email account or connectivity, and it would be delivered through snail mail.
How do multi-byte Unicode characters work over SMS? Languages that use an average of four bytes per character can only send 35 character messages? I'm not so sure "fair" is really part of the discussion. Twitter still has SMS turned on in some countries that use wide characters, right?
It's not hard at all to do this test yourself, and I've done it. Not sure what the argument is-- Twitter is fine with unicode characters. Are you seeing something different?
Maybe some of the clients or the SMS stuff doesn't like it, but the core site will let you type 140 unicode characters and share that with people.
There is no argument, I'd like to know how they handle it. Since I don't use a language that uses a lot of wide characters, nor do I use twitter via SMS (I barely use my twitter account as it is), I'm interested as to what the experience is.
I'm not sure that's true, at least for Twitter. I just tried tweeting 140 ♥s, and all 140 of them showed up. (And I'm sure my followers now feel very loved)
Found it on hacker news & created a Firefox Ubiquity command for it. Here it is http://➡.ws/⇷
(tinyarrows don't show up as links on twitterfox though.)
Though I do not have any problems with tiny arrows on Twitter/Firefox3 Some of my friends complain that they see blah blah instead of the right signs. And when clicked they get errors. Must be a problem with unicode support I guess & it will be a major drawback for the tinyarrows :(
I just twittered via twitterfox. I checked on my Twitter page & it shows up well for me. But a friend using Fx3 on window$ just complained there are problems viewing it. WC.
Thanks a lot for the clarification. If it's not a hassle, can you ask them how they read Twitter? Did they visit the website or get through a different reader?
Microsoft Windows 5.1 (Build 2600.xpsp_sp3_gdr.080814-1236: Service Pack 3)
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US) AppleWebKit/525.19 (KHTML, like Gecko) Chrome/1.0.154.48 Safari/525.19
Reading from the front page of Hacker News. The links work but all of the fancy pants unicode stuff shows up as square boxes. This was also a problem when http://unicodesnowmanforyou.com/ was making the rounds.
I'm almost certain it's just a font issue since the Unicode Snowman takes advantage of a font embedding feature specific to IE (and works properly there). I'm not sure anything can be done about this, short of pushing browsers to install fonts.
It doesn't work at all for me in Firefox 3 on Windows. I get the same response after clicking on all links (feed reader, HN frontpage, this page):
Page Load Error
Address Not Found
Firefox can't find the server at www.➡.ws.
This is the same trouble as with Unicode snowman domain.
In one discussion here on HN I figured out it's probably because of some security measure to prevent phishing. There are some Unicode characters that look like normal ASCII letters (for example in Cyrillic alphabet), so if Unicode worked in urls, you could create malicious sites that would look like real ones:
Bookmarklet for it. It just displays the link (no copy to the clipboard like tinyurl, but I don't actually know javascript.), but it's better than nothing, I guess...
However, I'm not convinced that mainstream users will get comfortable with it (because it looks "different"). This, plus the fact that enough problems can exist from creation/sharing/clicking that it's likely just not worth the hassle.
Maybe it's just me, but lack of compatibility is a high price to pay for a character (or two) shorter URLs.
It only beats our own ri.ms shrinker by three characters, but that's still something people might consider. And because it uses unicode, it will be a very long time before it moves beyond even 2 character suffixes.
I'm with you. Recently I've seen people Twittering about Twiddla with tinyurl'd links that are actually longer than the domain name. It's just silly.
I think that SEO is really making things worse for URLs. Remember 4 years ago, when "The URL was the New Command Line"? And it was considered a feature that you could look at a URL like http://site.com/user/3916 and hack it to get similar pages for other users? Can't do that with http://yoursite.com/this_was_an_article_that_i_wrote.
I built these a while back but they haven't seemed to pick up traction. str8.to uses google's "i'm feeling lucky" - vanity urls without having to go register them.
Works fine for most people on Linux, too. It's probably one of your addons or settings tweaked from default, see this thread and let me know if that's not the case?
48 comments
[ 0.21 ms ] story [ 103 ms ] threadFor example, Hacker News is:
http://➡.ws/퐐
They won't work in IE6, but that's just the way it goes. Should work great on Twitter, hopefully.
the whole point of a tiny URL is it being typable. what's with the symbols :|
For example, emails and http:// addresses are already recognized as such in places like the gnome terminal (which is great then for irssi IRC) or even in this forum comments. But can't just domain.com/that/ be enough info to ne recognized as an url?
www.➡.ws/퐐
http://➡.ws/퐐
The www version actually works on Twitter, so we may make an easy API for that one for tools to use.
Based on your error message, I bet it's not FF3 but whatever showed you the link. If I had to guess, you clicked on the link/page through a feed reader or something?
Can you confirm whether it happens when you click directly from this page/site on Hacker News? (which seems to be handling them great)
I'm using Firefox 3.0.5 on Windows XP.
FYI, the other link you gave in http://news.ycombinator.com/item?id=498246 too doesn't work for me: "Firefox can't find the server at www.%e0%b2%95%e0%b2%b0%e0%b3%8d%e0%b2%a8%e0%b2%be%e0%b2%9f%e0%b2%95.com."
http://news.ycombinator.com/item?id=498378
I'm pretty sure services like Twitter count bytes, not characters. So unicode will not give you much by ways of savings characters.
Also, analytics make the shorteners these days, not length. They're all pretty much short enough, no?
Sorry for the accidental downvote-- I have a reflex of downvoting people who reply to the wrong comment and I forgot this was my own thread. Sorry. :(
Twitter is doing a character count and doesn't care about bytes-- if it did, it means that tweeting in non-ASCII languages only gets half the characters. That's not exactly fair.
If people like the service, we'll be adding metrics and even allow people to search for tinyarrows that were associated with a particular domain.
It's more of something fun-- we're not trying to change the world or anything. We make a lot of sites/toys and this is just one of the fun ones to share with people.
At the peak of .Com days, there was a startup that promised to bridge e-mail and snail mail: one can use their service to email people who don't have an email account or connectivity, and it would be delivered through snail mail.
Maybe some of the clients or the SMS stuff doesn't like it, but the core site will let you type 140 unicode characters and share that with people.
http://en.wikipedia.org/wiki/SMS#Message_size
SMS can use the default 7-bit GSM encoding, an 8-bit encoding or UTF-16, giving maximum lengths of 160, 140 or 70 chars/code points, respectively.
That said, I think a lot of phones don't have full unicode support.
Some IM messangers and emails will mess IDN links up, so I'm on a crusade to notify vendors to get things working better.
Also these don't work in IE6. Are they using that?
Can they click on a page like this? http://ಕರ್ನಾಟಕ.com ?
Thanks in advance!
Are you clicking them from a feed reader? Or from the Hacker News page? What kind of error do you get?
Sometimes feed readers or the like urlencode the domain name and it confuses IDN resolvers in browsers.
If it's not working directly from http://news.ycombinator.com/item?id=498051 can you tell me what version of Chrome and Firefox you're using?
Thanks in advance!
Microsoft Windows 5.1 (Build 2600.xpsp_sp3_gdr.080814-1236: Service Pack 3)
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US) AppleWebKit/525.19 (KHTML, like Gecko) Chrome/1.0.154.48 Safari/525.19
Reading from the front page of Hacker News. The links work but all of the fancy pants unicode stuff shows up as square boxes. This was also a problem when http://unicodesnowmanforyou.com/ was making the rounds.
I'm almost certain it's just a font issue since the Unicode Snowman takes advantage of a font embedding feature specific to IE (and works properly there). I'm not sure anything can be done about this, short of pushing browsers to install fonts.
It works fine in my Chrome:
Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US) AppleWebKit/530.1 (KHTML, like Gecko) Chrome/2.0.166.1 Safari/530.1
I'll have to see if I can install the original Chrome 1.0 on another machine and check it out, too.
Thanks a lot for the feedback.
In one discussion here on HN I figured out it's probably because of some security measure to prevent phishing. There are some Unicode characters that look like normal ASCII letters (for example in Cyrillic alphabet), so if Unicode worked in urls, you could create malicious sites that would look like real ones:
http://EXAMPLE-ВANK.COM (fake)
http://EXAMPLE-BANK.COM (real)
Also, check your about:config and filter by IDN, you should see a lot of settings there. Any of them not set to default?
Especially check this one: network.enableIDN
by default it is set to True.
What other Firefox addons do you have that might be parsing URLs before you visit them? Any?
Thanks a lot for the feedback!
javascript:void(location.href='http://tinyarro.ws/api-create.php?url='+encodeURIComponent(l...)
Also, is there a better way to display this in the comments?
However, I'm not convinced that mainstream users will get comfortable with it (because it looks "different"). This, plus the fact that enough problems can exist from creation/sharing/clicking that it's likely just not worth the hassle.
Maybe it's just me, but lack of compatibility is a high price to pay for a character (or two) shorter URLs.
And the URLs are dramatically shorter than a character or two-- for example:
http://➡.ws/껻 (6 chars after protocol)
versus:
http://tinyurl.com/aqehn8 (16 chars after protocol)
http://ri.ms/9i0 (9 chars after protocol)
It only beats our own ri.ms shrinker by three characters, but that's still something people might consider. And because it uses unicode, it will be a very long time before it moves beyond even 2 character suffixes.
Anyway, thanks for the feedback. :)
I think that SEO is really making things worse for URLs. Remember 4 years ago, when "The URL was the New Command Line"? And it was considered a feature that you could look at a URL like http://site.com/user/3916 and hack it to get similar pages for other users? Can't do that with http://yoursite.com/this_was_an_article_that_i_wrote.
Seems like a step backwards.
I built these a while back but they haven't seemed to pick up traction. str8.to uses google's "i'm feeling lucky" - vanity urls without having to go register them.
http://news.ycombinator.com/item?id=498363
http://➡.ws/♲♲