Vorbis - Oct 2003 - Desirable Properties of a Real Metadata Format

I suggest gzipped XML. Feel free to suggest something that isn't XML if you
dislike it that much.

Anyhow, as a practical matter, Vorbis-style metadata lacks a Really Good Way
to have the same data in multiple languages. Or at least a widely-agreed
upon way. So, howzabout...

<data>
<title xml:lang='en'>Mars, the Bringer of War</title>
<title xml:lang='de'>Mars, der &#x00dc;berbringer des
Krieges</title>
<title xml:lang='fr'>Mars, qui apporte la guerre</title>
</data>

It's not just this /The Planets/ album that does this;
http://www.cdbaby.com/cd/chamberchoir has a number of weird things about its
track titles, including "13. I Think Upon the Faithfull [sic] Day" and
to a
slightly lesser extent, "10. The Cherubimic [sic] Hymn". At least if
the
original names were in the file...well, that'd be nice. I guess.

Does anyone have any idea how to deal with track num, er, indices like "9a,
9b", &c.? The best I can suggest is sorting numerically (i.e., not
lexicographically) and then allowing the inclusion of some sort of
comparison function...

Ew.
--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'vorbis-request@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

> I suggest gzipped XML. Feel free to suggest something that isn't XML if
you
> dislike it that much.
You cannot use XML for streamable metadata, at least not without
changing what you mean by XML.  I wish people would please stop bringing
this up.

jack.
--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'vorbis-request@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

On Thu, Oct 16, 2003 at 10:45:11PM -0700, Nathan I. Sharfi wrote:
| I suggest gzipped XML. Feel free to suggest something that isn't XML if
you
| dislike it that much.

That makes a good deal of sense to me.  I'm not normally an XML zealot,
but it makes a lot more sense to use XML than coming up with our own
structured text format.  Another possibility would be to use some kind
of s-expressions; XML has in its favour an abundance of parsers and
other tools to manipulate it.

(Although: after reading Jack's post and searching the list archives it
sounds as though people have suggested and rejected XML-based metadata
over three years ago...)

| Does anyone have any idea how to deal with track num, er, indices like
"9a,
| 9b", &c.? The best I can suggest is sorting numerically (i.e., not
| lexicographically) and then allowing the inclusion of some sort of
| comparison function...

No!  A better way is to treat consecutive digits as if they are base-ten
numbers, and sort any letters or other characters lexicographically.
For an example of the kind of thing I'm talking about, take a look at:

       
http://www.debian.org/doc/debian-policy/ch-controlfields.html#s-f-Version

(The bit I'm referring to begins "The strings are compared from left to
right.")

Cheers,

Cameron.

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'vorbis-request@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

Nathan I. Sharfi wrote:
> I suggest gzipped XML. Feel free to suggest something that isn't XML if
you
> dislike it that much.
> 
> Anyhow, as a practical matter, Vorbis-style metadata lacks a Really Good
Way
> to have the same data in multiple languages. Or at least a widely-agreed
> upon way. So, howzabout...
> 
> <data>
> <title xml:lang='en'>Mars, the Bringer of War</title>
> <title xml:lang='de'>Mars, der &#x00dc;berbringer des
Krieges</title>
> <title xml:lang='fr'>Mars, qui apporte la
guerre</title>
> </data>
> 
> It's not just this /The Planets/ album that does this;
> http://www.cdbaby.com/cd/chamberchoir has a number of weird things about
its
> track titles, including "13. I Think Upon the Faithfull [sic]
Day" and to a
> slightly lesser extent, "10. The Cherubimic [sic] Hymn". At least
if the
> original names were in the file...well, that'd be nice. I guess.
> 
> Does anyone have any idea how to deal with track num, er, indices like
"9a,
> 9b", &c.? The best I can suggest is sorting numerically (i.e., not
> lexicographically) and then allowing the inclusion of some sort of
> comparison function...
> 
> Ew.
On XML: I would recommend an XML-like format that doesn't include any 
root tags, so as to facilitate the cat file1.ogm >> file2.ogm 
functionality. Also, the metadata stream(s) would have to support 
multiple gzips, so as to prevent having to load too large of a cache and 
to allow for arbitrary splits. To the end of defining tags, I think of 
OGG largely for movies, and as such, one might envision the following 
entities:

<chaptermark />
<subtitle /> (could incorporate XHTML here, and even MathML if we were 
crazy enough... jk)
<subtitle forced="en, jp" /> <!-- forces under en and jp
languages- use
for pretend languages and such. -->
<oggmenu>
        <?ogg import-src="[url]" ?>
        <bgimg src="[url]" />
        <option src="[url]" href="[url]" />
        <option src="[url]" action="[script call]" />
</oggmenu>

Anyway, just a few thoughts...

-- 
Christopher Granade
   (cgranade@greens.org)
   Webmaster, Green Party of Alaska
   Secretary, RLC of Lathrop and Stevens Halls
   Secretary, RHA of UAF

<p>"Everyone has to find their own reason."
   -Cloud, Final Fantasy VII

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'vorbis-request@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

On Friday 17 October 2003 18:45, you wrote:
>  I suggest gzipped XML. Feel free to suggest something that isn't XML
if
> you dislike it that much.
Perhaps RDF,  as it's already a general framework for addressing metadata, 
serialized as XML (which is usually the case for RDF), and perhaps gziped, 
though it's not like the data will be so great compared to track size that 
this will be a problem (but you'd have to extract it from the ogg stream 
anyway, so why not, I suppose...)

However, seems a bit premature to start talking about technologies before 
discussing who wants what kind of metadata and why.

John
--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'vorbis-request@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

On Thu, 16 Oct 2003 22:45:11 -0700 (PDT)
"Nathan I. Sharfi" <nisharfi@csupomona.edu> wrote:
> Does anyone have any idea how to deal with track num, er, indices like
"9a,
> 9b", &c.? The best I can suggest is sorting numerically (i.e., not
> lexicographically) and then allowing the inclusion of some sort of
> comparison function...
I propose using two "tags"/values/whatever (no, not Vorbis tags).
Either order and name, order being a zero-based (or perhaps one-based) integer
and name being the "value" to show, which in this case would become:
(assuming zero-based)
8 9a
9 9b
10 9c
That is, if there are no other parted tracks like this before no. 9.

The other usage is more restrictive, using something like track and part where
part is an arbitrary string
and track is a one-based integer. It would handle this case, but perhaps not
others.

<p>Oscar
--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'vorbis-request@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.