[Playlist] Questions on spec about meaning of identifier and link

[Lucas Gonze]

I should give a little more detail on how a Musicbrainz track URI in an 
identifier element can be converted into a URL that can be dereferenced to 
return a song file.

Let's say you are an XSPF reader and you have been invoked to open a 
playlist which contains this XML fragment:
<track><identifier>http://musicbrainz.org/track/541f8f37-c9f3-4567-b234-9806d25ca433.html</identifier></track>

(A fancy name for a "reader" is "content resolver," but I'm going to use 
the wording "reader" here).

If you happened to support this identifier style, one way you might do it 
is like this.  You would have code which knew that a URI in the above 
style could be converted into a URI which could be dereferenced to get an 
XML file of metadata about the music being identified.    In this case the 
XML URI is:
http://musicbrainz.org/ws/1/track/541f8f37-c9f3-4567-b234-9806d25ca433?type=xml&inc=artist+releases

It appears that there is a track identifier shared between these two URIs, 
and you (the reader) might have code which took a URI in the <identifier> 
style, parsed out the shared portion, and inserted it into an XML URI.

You then parse the XML to get metadata, and you learn that the track title 
is "Mandy", the artist name is "Barry Manilow", and the album title is 
"Greatest Hits: The Platinum Collection."

Now that you have metadata, you can go get a dereferenceable URI for the 
song.  Your user might have told you already that they have a subscription 
to the Napster subscription service; if you supported that, you would 
probably use a web service provided by the subscription service to query 
for a URL, using the song metadata as query terms.  Another approach would 
be to construct a Magnet URI 
(http://en.wikipedia.org/wiki/Magnet:_URI_scheme) along these lines:
magnet:?kt="Mandy"+"Barry Manilow"+"Greatest Hits: The Platinum 
Collection."

I hope this all makes sense.  The identifier element is an important 
feature of XSPF, but it's subtle and poorly documented.  Please don't 
hesitate to ask for clarification on- or off- list.

-Lucas



On 6/13/07, Lucas Gonze <lucas.gonze at gmail.com> wrote:
> There is a subtle thing to realize about http:// URIs -- they can be
> just identifiers and not at all locators.  Because of this they have
> absorbed the function of URNs, which are now a subset of URIs (the
> subset which is just an identifier and not a locator).
>
> For example, let's say you wanted to have the recording of the song
> "Mandy" on the album "Greatest Hits: The Platinum Collection" by the
> artist "Barry Manilow" in your playlist, and you wanted your song
> reference to work regardless of where the listener happened to be able
> to fetch the song from.
>
> One thing you could do is go look it up on Musicbrainz.org and find
> the track page, then use the URI of the track page as the identifier.
> That URI happens to be at
> http://musicbrainz.org/track/541f8f37-c9f3-4567-b234-9806d25ca433.html
> , so the XSPF for this recording would be:
>
> 
<track><identifier>http://musicbrainz.org/track/541f8f37-c9f3-4567-b234-9806d25ca433.html</identifier></track>
>
> Notice that dereferencing this URI does not give you a copy of the
> song.  The URI is just an identifier, and not a locator.
>
> Also, it doesn't matter that this particular URI goes to a page of
> text and metadata.  The URI doesn't have to even be real.  It just has
> to be usable as an identifier which the publisher and content resolver
> can both understand.  For example, the domain example.com is forbidden
> (by the rules of DNS) to map to any real world web server, but the URI
> http://example.com/track/541f8f37-c9f3-4567-b234-9806d25ca433.html is
> perfectly legal in an XSPF identifier element.
>
> -L
>
> (Hey Tomas, nice to see you.  :)
>
> On 6/13/07, Tomas Franzn <tomas at lightheadsw.com> wrote:
> > On 13 jun 2007, at 22.12, Ross Mohn wrote:
> > > 1. Why does the spec indicate that a track identifier "MUST be a 
legal
> > > URI"? I would think of this element as specifying a CD id code or
> > > something of that sort, not a URI. Doesn't the phrase "MUST be a 
legal
> > > URI" conflict with the phrase "location-independent name"?
> >
> > Hi,
> >
> > Remember that URIs are not the same as URLs (the L stands for
> > Location). URIs can be URNs, meaning something like this (for ISBN
> > numbers):
> > urn:isbn:0-395-36341-1
> >
> > Note that the above does not specify a location. It's up the client
> > to figure that out.
> > Wikipedia has more info. http://en.wikipedia.org/wiki/URI
> >
> > (Hi everyone, BTW. It's been awhile since I last posted here.)
> >
> > Tomas Franzn
> > Lighthead Software
> > http://www.lightheadsw.com/
> >
> >
> > _______________________________________________
> > Playlist mailing list
> > Playlist at lists.musicbrainz.org
> > http://lists.musicbrainz.org/mailman/listinfo/playlist
> >
>