Re: ISSUE-27, was: Report on testing of the link relations registry

On 2010-08-17 10:31, Maciej Stachowiak wrote:
> In any case, the registry is now up and the test has been done. It
> doesn't seem like we will gain much more data from further
> investigation. So it seems to me the right thing to do is check
> whether anyone wants to propose a different solution than the IANA
> registry (or some variant of the IANA registry approach), given the
> information we now have available.

Given that the IANA registry seems to be a complete and utter failure 
that doesn't meet any of the requirements for HTML5, we should instead 
develop a simplified registration system that does, but which also 
improves upon the wiki to resolve its technical limitations.

I think the requirements for such a system should include:

* The ability for anyone to submit a new relationship, and for it to
   be granted the status of a *proposal*.  The information required for
   proposals must be clearly and objectively defined, with a low entry
   barrier.  (e.g. relationship name; description; effect on <a>, <link>
   and <area>; optional link to more details; and optional synonyms)

* Some basic moderation process to block obvious spam, but any clear
   attempt at proposing a legitimate relationship should be accepted.

* The ability to edit and update existing relationship entries, with
   moderation only to prevent spam.

* A clear and simple process and relatively low entry barrier for a
   proposal to be upgraded from *proposal* to *accepted*. (e.g.
   Following the Microformats process, or published in a spec by the W3C)

* A clear and simple process for a submission to be withdrawn by the
   applicant, or rejected after following said process.

* The ability to specify the relationship as either a *Hyperlink* or
   *External Resource*

* The ability to allow it as conforming on either only <link>, or only
   <a> and <area>, or either.

* Multiple output formats to query the registry. (e.g. human readable
   HTML and/or plain text, and machine readable JSON, etc.)

Ideally, there would be a simple web interface for submitting and 
updating entries.  The existing wiki does meet most of those 
requirements, except the available output formats, and it's not the most 
ideal interface for the job, and doesn't have the best moderation 
system.  So any new system we adopt must improve upon the existing wiki 
based approach.

Lachlan Hunt - Opera Software

Received on Tuesday, 17 August 2010 09:52:39 UTC