Re: DC metadata

Henri Sivonen On 09-05-25 16.52:
> On May 25, 2009, at 17:36, Philip Taylor wrote:
>
>> http://philip.html5.org/data/meta-names.txt - looks like thousands of 
>> people use <meta name="DC.*">.
>>
>> So, people don't seem to follow the profile requirement much.
>
>
> In interesting question is:
> What existing consumers (other than the one Julian wrote) would break 
> if the dc.* tokens were pre-defined with such that the prefix is 
> always 'dc'?

(1) Consumers and producers. Both need to be able to discern between the 
two profiles that  DC now has.

(2) Dublin Core has been designed as a specification that can be used to 
also link to other meta data conventions. It is as open ended as RDFa is 
in that sense - anyone can use @rel="schema.*" to define a "DC 
namespace" - an Philip's research shows that it is being done [1]. So 
just as with  RDFa, it also possible to mix namespaces when using DC. 
Asking them to require "DC" is to ask them to limit their specification.

(3) Personally, I wonder why RDFa was based on xmlns:*= rather than on 
e.g. <link rel="schema.prefix" ....> or  e.g. <link rel="xmlns:prefix 
... > since using <link@rel> instead of <html@xmlns> would have worked 
fine in HTML 4 as well. (Note how the 2008 profile gives both XHTML and 
HTML examples! [2])

I think the answer is found in what  Manu said about link relations 
being a done deal w.r.t. HTML 4 (and XHTML 1?) [3]. It seems relatively 
clear that there would have been less use for @profile if there was a 
common register for link relations and if the "default profile" of HTML 
(in any of HTML/XHTML's flavours) was possible to expand in a parallel 
process to how HTML/XHTML itself develops. Why go looking for the 
another HTML specification if all one need is a new link relation or 
other profile governed aspect?

So far, no one has proposed @schema.* in the WHATwg wiki or anywhere 
else. And I have no clue about whether the Dublin Core Metadata 
Initiative would have registered it or not. After all, it is is their 
extension. And may be here we see another feature of the profile 
concept: It formalises a way that a party outside the W3 can specify 
some aspect of HTML.

PS: I find it problematic that the WHATwg wiki contains only proposed 
link relations. Whereas those link relations that have made it to the 
draft are not represented there. It becomes less useful as a reference 
that way.

[1] http://philip.html5.org/data/link-rel-rev.txt
[2] http://dublincore.org/documents/2008/08/04/dc-html/#sect-3
[3] http://www.w3.org/mid/4A1A8A99.2040501@gmx.de
-- 
leif halvard silli

Received on Monday, 25 May 2009 16:04:45 UTC