View Full Version : Re: Multiple Artists and Genres

Guy Albertelli II
2004-03-10, 13:07
Hi, Pat,

> > > This is why I think the whole "tag" vision is already at a dead end.
> > > We have to move the classification of out the file structure.
> >Why? I _really_ like having the metadata about my data associated with
> >my data. That way when I move my data around it's still there.
> If you have something that looks like a relation of data, you have to
> have a relational structure.

I'd argue that you don't have to. The data is useful even immaterial of
whether the relation is specified.

Knowing the Composer(s) is useful whether or not you know how it
relates to anything else as it is a fact.

Genre starts crossing the line from fact, to opinion which is why it
isn't as useful devoid of context.

> I have no complaint about where it is kept. But unless you can define,
> store, and access the data in the relation, and not in some implied
> flattened structure, then you can't process the data. Computers are
> good with relational data.

But it's not in 'some' implied structure. It is in the ID3V2 implied
structure. As with any standard, some people do a good job on
implemting it and some don't. No new 'standard' of exchaning data
will fix that.

I guess if you flipped to XML attached to the end of the MP3s then
you'd have an expressed relation. Of course we likely end up with 30
different standards then.

> I'm not suggesting that we force everyone to not use tags,
> flat structures or anything else. But I equally don't want
> to try to cram an essentially relational structure into a flat model.
> Genres are just one example. Song flow is another. Sometimes
> you want separate songs, sometimes they have to seque together.
> Movements in classical work, songs in a rock "concept" album, etc.
> are meant to be played together. You can't store that information
> in a tag for one song.

Acutally, ID3 has that too TIT1, TIT2, TIT3 are to be used for noting
title(s) at various levels that a work belongs too.

> My vision is that we stop thinking about "songs" and think about
> collections of music. Then mood, tempo, style, period, and flow
> can be managed. In the olden days, good radio stations did this,
> as did Muzak.

I really like being able to give a friend an mp3 and know that he know
a large number of facts about the song.

I can't see any really better way of doing this. If all of the info is
_only_ stored in the database then I need to generate some kind of
report when I want to give him the file. And that report is inherently
a flattened stucture of the data available. I don't see how ID3v2 is
really all that bad at being an extraction of data for the friend,
since the standard is fairly widely and generally not badly

guy (AT) albertelli (DOT) com LON-CAPA Developer 0-7-4-7-