RE: 3023 update (was Re: Agenda TAG Telcon: 8th Nov 2004)

This is the same case in XML as well. Tim is basically implying that every XML vocabulary has a distinct MIME type and file extension which is simply not the case in the wild. For example, the RSS files usually served with a ".xml" extension are typically RSS 0.91/RSS 2.0 files which have no registered MIME type. So the only thing that one validly serve them as is as */xml of some form or the other. In fact the author of the RSS 0.91/RSS 2.0 family of specifications advised serving them as text/xml as a best practice[0]. 
 
[0] http://blogs.law.harvard.edu/crimson1/2004/05/06#a1519
 
-- 
PITHY WORDS OF WISDOM
There is nothing more satisfying that having someone take a shot at you, and miss. 

________________________________

From: www-tag-request@w3.org on behalf of Dan Brickley
Sent: Sun 11/21/2004 12:54 AM
To: Tim Bray
Cc: www-tag@w3.org
Subject: Re: 3023 update (was Re: Agenda TAG Telcon: 8th Nov 2004)




* Tim Bray <tbray@textuality.com> [2004-11-20 19:35-0800]
>
> On Nov 20, 2004, at 3:00 PM, Dare Obasanjo wrote:
>
> >That is a very significant generalization you've made there.
> >Considering that both IIS and Apache default to serving files that end
> >in ".xml" as some */xml MIME type are you claiming that they are both
> >buggy or that every time someone puts a file that ends in ".xml" on
> >the Web it is a bug?
>
> I think that a file whose extension is .xml is usually evidence of a
> bug, yes.  Because the file is usually something else (often RSS), and
> the .xml is hiding that. -Tim

Oh, interesting. This shakes out slightly differently in RDF, where often
we're never sure if it's a "Dublin Core", "FOAF", "Creative Commons",
"DOAP,"Wordnet", "MusicBrainz" or "FOAF" file, since those
vocabs/namespaces only define dictionaries of terms that can get mixed t
ogether. The mixing is a blessing and a curse. RSS1 is a little different
'cos the spec (just about...) defines a document class too...

So RDF files are typically served application/rdf+xml though often
they have a "dominant namespace" whose presence drives the data
structures, with other namespaces typically being somewhat
annotational in their use.

Dan

Received on Sunday, 21 November 2004 12:26:20 UTC