I have a few things to say about the
canonicallink relation is defined to be same-origin. That does not work well for what people want from
canonicallink relation is about resource equivalence, not URL equivalence. I.e. that the document you are looking at is similar to this other document, not that this URL redirects you to this document.
revwas not removed from HTML5 as a joke. It is frequently misunderstood/misspelled and can always be expressed as
Just call the link relation
shorturl, register it, and be done with it.
I completely agree. The distinction between rel and rev was already confusing Slashdot readers, so Chris Shiflett had to add a note to his article to explain.
After seeing some running code and actual examples, #2 in your list bothers me the most. Also, the fair number of typos of both
rev I've seen throughout this discussion is kinda troublesome. (I don't see
rel misspelled as much in the wild, so yay for cut and paste.
I'm still not sure that
shorturl is a great
rel value (it is also the name of a company providing a shortening service) but I've been losing interest as the discovery process turned bikeshed painting into an extreme sport.
Seems like this is already done: rel="shortlink"
I blogged briefly about this myself and I totally agree with #2 on your list.
To my mind, applying a relation on a link that 301 redirects back to the original page is essentially self-referential. You may as well do
link rev="canonical" href="#" it would make as much sense!
I'm surprised the rev=canonical guys are still banging on about this... nobody outside of the web developer community has a good word to say about it and it's been universally criticised by the standards community (with good reason).
rev=canonical is saying "I'm the canonical URL and that URL over there points at me". That means it must only ever be used on the canonical URL itself - too bad for an infinite number of potential permutations. Then there's the likes of Matt Cutts pointing out that one should give, not take, canonical-ness but Chris Shiflett (one of the primary promoters) foolishly dismisses this feedback as "irrelevant". Mark Nottingham is more direct in Counting the ways that rev="canonical" hurts the Web but the rev=canonical fanboys cite this as "evidence that those writing the standards are going off track".
I suggest rel="shortlink" as an unambiguous solution to this fiasco - the short[_- ]?ur[il] option has more permutations than can easily be counted and delivers no additional value.
In order to discourage people from supporting many variations I'm now serving up warnings and errors when rev=canonical and rel=short*ur? are discovered at http://rel-shortlink.appspot.com/.
Sam your understanding of "canonical" is completely incorrect (it's ok though, it seems like there is a huge opposition that is making the same naive mistake). rev=canonical does not have to be restricted to being used on the canonical url itself, because if it is not the canonical then it will have rel=canonical (but doesn't have to), and rev=canonical actually is saying is "my canonical represents this" (here is other way to think of rev=canonical: "this document's canonical represents that"), and what rel=canonical is actually saying is: "that is my (this document's) canonical" (thus it can be used on the canonical url itself).
Furthermore, your understanding of rev seems to be off the mark in this instance, under your understanding of rev=canonical it would be saying: "this document is the canonical" and "this document's canonical is represents that". That is utterly wrong. It is impossible for the rev attribute to say "this document is the (as in the one and only) canonical for that", it could be interpreted as "this document is a canonical for that" (as you do) but that breaks the definition of "canonical", and in light of the def of canonical, "this document's canonical represents that" is the only interpretation that makes sense.
I know that many will be tempted to blame the confusion on the rev attribute, but really you all get rev, it's when a value comes in to play that you don't understand, like 'canonical' for instance.
Just because you are confused does not mean your solution is better.
Please read my deeper explanation here: Counting the ways rev="canonical" helps the Web and a rel="short*" rebuttal. It rebuttals the rest of your comment.
I apologize Sam, where I say "(as you do)", that is incorrect.
We agree that rev=canonical cannot be allowed to be interpreted as you described though, I just see a interpretation of rev=canonical that is more accurate, that I would like to hear rebuttals for, instead of rebuttals for the interpretation you but forward, because we agree there. And Matt Cutts only point against rev=canonical is this point you and I agree on.
Sigh, apologies to Matt now, his argument is more than just the point I mentioned in the last comment, but I can take care of the rest of his argument by saying simply verify rev=canonical by checking for a rel=canonical.
Erik, trying to redefine the
canonical link relationship does not make your case stronger.
Anne, I haven't redefined the canonical link relationship, you got the definition wrong was my point..
Did you read my blog post? I explain the true definition of canonical there, where I provide a rationale for the definition, unlike here, where you just state your definition, without a rationale..
@Erik: I tend to agree with Anne on this - your version of reality differs from the rest of us: "rev" is the reverse of "rel" and in any case it doesn't exist in the future and presents an SEO death trap for clueless webmasters... 'nuff said.
Regarding shorturl, it gives no advantage over shortlink and yet is confusing and has various intellectual property issues as well (copyright protected specification, trademark status asserted by shorturl.com, patent status unknown).
We're all trying to achieve the same thing here so I'm not sure what's with all the bikeshed painting - rel=shortlink is the only proposal lacking serious issues and yet you're still carrying on about rev=canonical?