Comments on Sam Johnston: rev=canonical considered harmful (complete with se...

Hey thanks for the heads up re: php.net - as at ri...

2009-07-06T02:02:24.913+02:00

Hey thanks for the heads up re: php.net - as at right now they have rev="canonical" rel="self alternate shorter shorturl shortlink".

So we already know why rev="canonical" is a clusterf--k but what about the rest of them?

rel="self" is defined by Atom (RFC4287) as "the preferred URI for retrieving Atom Feed Documents representing this Atom feed". Clearly that is far more likely to be the canonical URL rather than some redirector to it.

rel="alternate" signifies "an alternate version of the resource" (e.g. PDF, plain text), NOT the same thing.

rel="shorter" implies that the resource itself is shorter (e.g. an abstract or short video) as relations refer to the resource, not the URLs.

And then there's the only half sensible alternative, rel="shorturl", but that has many permutations (/short[_- ]ur[li]/) and actually started life as short_url.

rel="shortlink" is obvious, self documenting and virtually impossible to get wrong.

Sam

I've temporarily put aside the rev=canonical o...

2009-07-05T16:59:59.869+02:00

I've temporarily put aside the rev=canonical on my site ( http://free.naplesplus.us ) and am sticking with the rel="shorturl shortlink ...etc etc" basing it upon php.net's idea but without the rev=canonical.

I hear the for and against but until I get a better say-so from bigger authorities, I'm going to wait and see.

Kenneth Udut

The concept is sound, but the implementation is po...

2009-04-13T12:06:00.000+02:00

The concept is sound, but the implementation is potentially extremely dangerous - a single character slip (rev vs rel) could knock an entire site off its perch. I think the thing that really got me was the "another 30 minutes or less" production gloat - it took me less than 30 minutes to grok the implications. While I appreciate that we live in times of accelerating change some amount of application of the precautionary principle couldn't hurt.

Another thing I found particularly disconcerting about the completely unjustified perceived urgency of the situation was production deployments to sites like ars.technica, apparently by individuals who are quicker to engage their fingers than their noggins, without adequate change control and on a holiday weekend no less. Do we know that Google aren't going to blow up and dump the site at the next google dance? Yahoo? No, of course we don't but we could certainly have afforded to wait and see.

Sam

Ugh, and of course blogger gets my OpenID wrong.

2009-04-13T08:14:00.000+02:00

Ugh, and of course blogger gets my OpenID wrong.

No offense, but belittling rev="canonical" propone...

2009-04-13T08:13:00.000+02:00

No offense, but belittling rev="canonical" proponents and running code are probably not effective ways to sway people or convince them to work with you. I'd suggest you spend some more time distilling the argument against without insulting people or their work.

Or better yet, forget the syntax and spend time exploring the implications of the actual concept regardless of expression in HTML.