RDF/XML's declared media type in cheap hosting environments

Phil Archer wrote to <mailto:www-rdf-interest@w3.org> on 1 February 2005 
in "Does it matter? (was Re: .rdf file extension security?)" 

> Clearly there are a lot of servers in use that don't support the RDF 
> MIME type by default.

Whether you mean installations ("the server listening on port 80 at 
www.example.com") or the software used in installations (for example, 
Apache 2.0.48), I would agree.

> However, they do support .xml.

Are we discussing installations or the software that powers them? And 
how do they support the mapping of the file-name extension ".xml" to 
Internet media types (a.k.a. MIME types, content types)? Do they map the 
".xml" extension to the media type "application/xml"? Do they map the 
".xml" extension to the media type "text/xml"? If the latter, do they 
label the character encoding scheme at the HTTP level by copying the 
encoding given in the XML declaration?

> My organisation's 
> aim is to get a significant number of websites to include RDF 
> descriptions of their content for PICS-like (child protection) reasons.

That seems a difficult goal. Good luck.

> See http://www.w3.org/2004/12/q/doc/rdf-contentlabels.html

In section 2.2, "Use case 2: Distributed content production, distributed 
label control":

The "link" elements have "href" attributes whose URI-refs have fragment 
identifiers that begin with digits. Such fragment identifiers never 
occur in conformant RDF/XML, so the URI-refs are poor examples.

I wanted to point to the examples or to their section with a URI-ref, 
but the only identified fragments in the document are empty "a" 
elements. More appropriate would be "div" elements with "id" attributes 
and with content. A sprinkling of semantics in today's Web can only help 
the Semantic Web. To take a slightly different tack, I paraphrase from a 
scrawling in my beloved projection booth:

“ ‘Dolby Digital SurroundSound’? Who gives a shit? I want ‘focus’!”

> Whilst it is easy for a professional to add a MIME type to a server, for 
> hobby webmasters (and, let's be honest, a lot of "professional web 
> designers") this is well beyond what can be expected.

You seem to assert that the task is too difficult for ordinary folk to 
handle, an assertion with which I disagree. I assert that the problem is 
sloth and apathy.

Astonishingly, in sites that have the technical capacity and social 
motivation to do things properly, one can still find RDF/XML served as 
an incorrect media type. Consider, for example, the Dublin Core Metadata 
Initiative's definition of its elements 

Date: Tue, 01 Feb 2005 21:45:22 GMT
Server: Apache/1.3.28 (Unix) mod_jk/1.1.0
Last-Modified: Mon, 20 Dec 2004 17:38:12 GMT
Etag: "4400b8-3ad8-41c70e04"
Accept-Ranges: bytes
Content-Length: 15064
Keep-Alive: timeout=15, max=100
Connection: Keep-Alive
Content-Type: text/plain

> So, the question is, does it make a practical difference if an RDF/XML 
> instance is in a file with a .xml extension and a MIME type of 
> application/xml, rather that nicely in a file with a .rdf extension and 
> a MIME type of application.rdf+xml?

Let me put it this way: would you accept the presentation of the rating 
metadata as, for example, a jumble of less-than signs and greater-than 
signs in Microsoft Internet Explorer? That is one of the consequences 
that a person invites by serving what should be RDF/XML as 
"application/xml". An incorrect media-type label does further damage in 
that it essentially hides RDF from scutters. Or forces the scutters into 
wasteful misbehavior, checking every XML resource to sniff for RDF.

You could take your issue to the W3C's evangelism list 
(<mailto:public-evangelism@w3.org>). However, the direct approach is to 
contact the people who write the server software. Consider Tim Bray's 
success with the Atom format, detailed in "Tribal Drumbeat" 

Etan Wexler.
“I’m not gonna say I’ve had my way. You sure saw to that. But what else 
can I do?”
   —Dinosaur Jr., “Not the Same”

Received on Tuesday, 1 February 2005 22:19:31 UTC