Re: ISSUE-53: mediatypereg - suggest closing on 2009-09-03

On Mon, 31 Aug 2009 13:22:41 +0200, Julian Reschke <>  
> Anne van Kesteren wrote:
>> On Mon, 31 Aug 2009 13:02:08 +0200, Julian Reschke  
>> <> wrote:
>>> Anne van Kesteren wrote:
>>>> ...
>>>> I continue to wonder what I'm missing here. Is this a requirement of  
>>>> media type registrations? If so, do you have a pointer?
>>>>  Furthermore, if this is a requirement, why are references from a  
>>>> non-normative section sufficient?
>>>> ...
>>> Please elaborate: which non-normative section?
>> only has references to older HTML  
>> versions in the non-normative introductory section.
>>  Is that sufficient to able to answer my questions?
> No. Anything in an RFC is normative unless it's explicitly stated  
> otherwise.

I see. Earlier this month you did not disagree with me on this:

Can you answer my questions assuming the last sentence is dropped. That  
is, why is this a requirement for media type registrations and where is  
that documented?

>>>> Apart from this whether this is or is not a requirement, what is  
>>>> useful about this being defined in HTML5 if it has absolutely no  
>>>> effect on anyone whatsoever?
>>> It isn't. It was Ian's choice to do it this way. My proposal is and  
>>> was to leave the registration in a separate document, which can  
>>> continue to also reference previous specs.
>> That is not an answer to my question. But since you put it this way,  
>> why would the media type registration document have to reference the  
>> previous specifications?
> Because the point of a media type registration is to point recipients to  
> a description of the format, sufficient to understand the document.

Is this documented somewhere?

And in what way is HTML5 not sufficient to understand older documents? Do  
you think it would be better if UAs used SGML parsers for non-HTML5  
documents and leave it undefined as to when they should invoke them for a  
text/html byte stream?

>> There are multiple versions of XML 1.0, only a single one is  
>> referenced. What does that imply?
> It implies that when RFC 3023 gets revised, the reference will need to  
> be updated. Note, btw, that it uses the un-dated URI as reference.

Should it only point to the latest version or all five?

Anne van Kesteren

Received on Monday, 31 August 2009 11:35:59 UTC