Re: Report on testing of the link relations registry

On Sep 1, 2010, at 14:37 , Ian Hickson wrote:

> On Wed, 1 Sep 2010, David Singer wrote:
>> A formal registry is a place where you can go to find out what is 
>> actually happening
> Very few of the Web-related registries fit this criteria. For example, the 
> Microformats registry has "pingback" in it; the rel="" registry does not 
> (and the application was rejected, despite the keyword being in very wide 
> use). The MIME types registry still doesn't have image/svg+xml, despite it 
> being a ten-year-old type.

Then it's an informal type, and someone else could register it with a new, conflicting meaning.  I can't speak to the history of other registrations.  Generally, an RA should accept properly documented and formatted registrations; it is the job of the person or organization registering to make sure that the code point is, in fact, useful, makes sense etc.

> Formal registries, at least as implemented so 
> far for the Web, have been a disaster in terms of how well they reflect 
> reality.

Mostly, I fear, because people don't seem to go to the trouble of using them...and then complain when what they want to know isn't in the registry or cannot be found.

> The microformats registry is far more up to date than the link 
> relations registry. There's no reason we shouldn't consider it the 
> official place to look to see what a link relation's spec is, or to ensure 
> that we aren't overlapping with someone else when we invent a new type.

But the 'we' in that sense is anyone who can edit the Wiki...which is anyone, as far as I can tell.  It seems there is no sanity check.

IANA is very successful, respected, and useful, for many many types.  It is a formal place that handles registration.  Since the people setting up the registry have indicated that they want to make it work, let's run with that and not re-invent something.

David Singer
Multimedia and Software Standards, Apple Inc.

Received on Wednesday, 1 September 2010 22:04:10 UTC