The only truly safe option that doesn’t require people to cover multiple bases is rel=shortlink.

Sam

But whith rel=”shortcut” there’s another problem: There are many pages that define their bookmark icon via rel=”shortcut icon”. Since the rel attribut is a space separated list, the shortcut icon would be your short url.

Using short or shorter is even not that correct, because one would expect a shorter version of the content.

And putting alternate in the list is even worse because the short url is not really an alternate to view the content since it just redirects.

I figured out rel=”shortcut redirect” would fit best.

Sam

I’d suggest that rel=”short” is the appropriate alternative (shorter? shorter than what? is that guaranteed to be the case?)

Sam

If a url A1 can claim it is the canonical url for another url A2 on the domain A, that opens up the possibility of hijacking attacks—especially on freehosts. That’s why when my team at Google built consensus for rel=canonical, we said that urls could only give away canonical-ness, not take it from other urls. Splatting canonical-ness forward from a url is safe, but claiming canonical-ness from other urls opens up the possibility of attacks.

I smell another post coming from me in the near future that concisely states what I was trying to say here.

I am also bothered by the fact that this feature is being included in the markup. It really just feels like a protocol type of issue to me. Can someone explain why this is being considered for implementation in markup rather than as an extension to HTTP? HTTP does allow extensions, doesn’t it?

1. What domain it leads to. Iwant to see information in order to make an informed decision how safe and informative the resource might be.

2. If I’ve seen the page before. I basically stopped clicking links on Twitter, since I often find myself on a page where I’ve been before, following a different URL.

3. What the page is about. Today many sites have very well designed URL’s that in themselves are informative. Short URL’s make us loose that information.

I’m seeing shortened URL’s all the time now, even when the full URL would fit with margin within Twitters 140 character limit.

So this is my idea. Except for badly designed .net sites which tend to have URL’s that reads like a book, stop using these (beep) URL shorteners!

There are other things in HTML that are much, much harder to grok than rev. People seem to do okay. And certainly the early adopters implementing rev=”canonical” right now know what they’re doing.

If the argument “but people might use it incorrectly” were a valid concern, we shouldn’t even be talking about new HTML elements like header, footer, section, article, etc., because some people are almost certain to implement them incorrectly. But that doesn’t stop them being semantically useful constructs that should be ought to be in the spec.

Just like rev.

CSS Stylesheets are linked and can be grouped by @title on the link element and can be marked as preferred by adding @title and omitting ‘alternate’ on @rel. Thus all links with @rel=alternate would be secondary to the @rev=canonical link that has @title but omits @rel=alternate
See the HTML 4.01 spec for further information on linking persistent, preferred and alternate stylesheets.
http://www.w3.org/TR/REC-html40/present/styles.html#h-14.3.1

1. The debate about rev versus rel has to do with whether the former is going to be included in HTML 5. (HTTP is a closed spec and is unrelated.)

2. No one cares if lots of people continue to use third-party services. This idea isn’t for them.

3. Answering your last question, you are indeed looking at the canonical URL. This is the distinction between rev and rel; rev indicates a reverse relationship, so the other alternative is not the canonical one. (It’s the short one.)

Hope that helps.

When did we deprecate the HTTP redirect in favor of markup? If there is a canonical URL for this resource, then why am I not looking at it?