Re: "index" link relation

Julian Reschke, Thu, 23 Jun 2011 23:15:19 +0200:
> On 2011-06-23 23:03, Tantek Çelik wrote:
>> On Thu, Jun 23, 2011 at 13:47, Leif Halvard Silli wrote:
>>> Tantek Çelik, Thu, 23 Jun 2011 12:03:59 -0700:
>>>> I personally am not opposed to 'index' in particular (I've used it in
>>>> the past).
>>>> 
>>>> However, I strongly prefer that we follow at least some sort of
>>>> rational/scientific methodology in such iterations so as to provide
>>>> objective (repeatable) reasoning of our actions, decisions, changes.
>>>> 
>>>> So far I've been using the data available to reason how to treat
>>>> existing or previous rel values.
>>>> 
>>>> In short:
>>>> * if a rel value was in a draft and is missing (without explanation)
>>>> from the final spec, or
>>>> * if a rel value was in a previous version of and is missing (without
>>>> explanation) from an update to the specification (even a draft update)
>>> 
>>> A repeatable, objective criteria: HTML5 doesn't per se decide what goes
>>> into the Microformat registry. Rather, it is the opposite way. The
>>> Microformats registry is supposed to be the one which forms the basis
>>> for whether a link relation may pass the door to the HTML5
>>> specification.
>> 
>> cite/link to HTML5 spec text or WG decision text that supports this
>> "supposed to be" assertion?
> 
> That's usually the point of a registry.

+1

> As such, I disagree with what Leif said as well: "The Microformats 
> registry is supposed to be the one which forms the basis for whether 
> a link relation may pass the door to the HTML5 specification."
> 
> No! Usually the point of a registry is to *decouple* the container 
> format from the definitions of extensions. Once you have a working 
> registry, you don't need to include values into the base spec, except 
> for those which *need* to be defined there.

The rel-ownership decision states: "Defer to the Microformats community 
for cataloging HTML rel values". However, I also had in mind the WHATwg 
wiki, which I think was meant as a "sandbox" for relations which when 
accepted could - but peraps did not have to - be taken in to HTML5 
proper.

> For instance, HTTP does have a registry for status codes. We don't 
> *need* to include them into the base spec, unless they are somehow 
> "special".

I agree that it ought to be enough that the relation is accepted by the 
registry.
-- 
Leif H Silli

Received on Thursday, 23 June 2011 23:36:33 UTC