Raw

Update (2005-07-23) see : mod_enclosure

Suzan Foster has a schema ( announcement) for media "enclosures" with RSS 1.0 feeds. It can be used as follows:

If this were the RSS 2.0 :

<item>
   ...
   <enclosure type="foo/bar" length="65536"
url="http://foo.bar/baz"/>
</item>

the RSS 1.0 would be:

<rss:item>
   ...
   <enc:enclosure>
     <enc:Enclosure>
       <enc:type>foo/bar</enc:type>
       <enc:length>65536</enc:length>
       <enc:url>http://foo.bar/baz</enc:url>
     </enc:Enclosure>
   </enc:enclosure>
</rss:item>

Note that the length and type are optional - you can add just as much or as little data as you like.

Why might you want to use this rather than RSS 2.0's enclosure?

Marc Canter asks : Where's the meta data?

I wanna know who's on the recording, what are the subjects and topics of the discussion, where particular juicy quotes reside and most of all the length and Creative Commons license info on the audio.

All available using this module with RSS 1.0 and any other standard RDF vocabularies you may need - the particularly juicy quotes can be pointed to using HTTP Range markers (see Jon Udell's post for details) , and those can be described just like any other resources ( <dc:description>Juicy!</dc:description>).

The module can be used now, and is directly compatible with general metada vocabularies such as Dublin Core (for title, creator and so on), Creative Commons and other media-specific vocabularies such as MusicBrainz.

Dave Beckett's Raptor parser toolkit already supports this module (along with RSS tag soup, Atom 0.3 as well as full support for all the RDF vocabs like FOAF, RSS 1.0, Dublin Core, DOAP and OWL - announcement).

This goes way beyond the support offered by RSS 2.0 "extensions". Take the Raptor implementation. This can be used alongside the Rasqal RDF Query Library enabling rich querying of the song's metadata, potentially hooked into any other RDF available on the Web, such as MusicBrainz descriptions.

Very cool.

PS. I got a mail from Marc in which he says he'll tell Messrs. Curry and Winer they're just wrong (tongue meet cheek). I don't personally think they're wrong as such, just that they're missing a lot of easy tricks by stubbornly avoiding RDF.

For example, you could add extra metadata like Marc suggested to either format:

RSS 2.0 -

<enclosure type="foo/bar" length="65536"
url="http://foo.bar/baz" dc:description="Juicy!" />
...

RSS 1.0 -

<enc:Enclosure>
   <dc:description>Juicy!</dc:description>
...

Cue scratched-record…wikiwikiwiki…

The difference is that to support the first version extra work will be needed from every aggregator developer. As it stands, the added information is meaningless. On the other hand, applications that have RDF support already know that the second version describes a characteristic of the Enclosure resource. Given how widely-used <dc:description> is, chances are they'll have some idea how to render that information too. Thing is, when you only look at one scenario, one extension, it seems a lot easier just to use straight XML. But even in Marc's sentence above there are at least 4 extension terms called for (all of which are already covered in existing RDF vocabularies), and the extra work multiplies and the interoperability will decline as ball-of-mud add-ons are made. With RDF you have a framework designed for describing resources. It can make life a lot easier.

…last night a DJ saved my life…I got the pow-er…I wanna I wanna I wanna zigazig ah…payump up the jarm, pumpit pumpit pumpit…repeat to fade…

It's good that people are publishing the data.

It is also straightforward extracting whatever metadata is provided in an RSS 2.0 feed from any extensions, as long as it's provided in a consistent fashion (that part can't be taken for granted). Bung it through XSLT to RDF/XML, and you've got it. But given that RSS (all kinds) are essentially metadata formats anyhow, it seems a waste not to be explicit about it in the first place.

PPS. Why might you want to use this rather than RSS 2.0's enclosure? [Part 2]

So I can issue the following request:

Give me all the mp3 format audio files published by any of the people I know that work for Microsoft in the last week, that are less than 10 minutes long with a review rating better than 8/10, unless the artist name is "Happy Mondays" in which case just give them to me, my kitten's got a thing about Bez….

(anyone want to try that in Sparql?)