[Amarok] b0503f5 Move Playlists back into Meta.

[Bart Cerneels]

On Sun, Mar 28, 2010 at 17:33, Maximilian Kossick
<maximilian.kossick at googlemail.com> wrote:
> Could you please outline your arguments why playlists are a
> fundamental Meta type?
>
> Because all the other Meta types represent metadata describing audio data.
>
> Playlists are nothing of that sort, they are simply a grouping of
> individual audio files. They do not additional information required
> for understanding what a particular piece of audio data is.
>
> I don't think anybody is arguing that playlists are not a fundamental
> concept on their own. That's why they are in core/. But I do not see
> any reason for them to be in the Meta namespace instead of in the
> Playlists namespace (which is what that commit is all about).
>
> Cheers,
> Max
>

Let me base this one similarities with Meta::Track.

* They are also made available by a provider (ref. TrackProvider). It
can be part of a Collection consisting of a TrackProvider,
UserPlaylistProvider and QueryMaker.

* have different implementations (Sql, File, UMS),...
* Playlist is a fundamental type in the sense of track being the
fundamental type that has Album, Artist, Genre, ... as it's fields.
Playlist has only has one complex type field: TrackList. All the rest
are strings.

* Playlist could be queried on in QueryMaker as a new QueryType. That
way we can search for playlists that have tracks with certain albums,
artists, Genre, rating, ... in them.

I'm sure there are more arguments to be made, but the biggest one is
more abstract:
Playlists should not be treated as a second class citizen in Amarok by
keeping them out of the true hard of amarok's core: Meta and
Collection. It's as much a technical issue: Playlists not supported by
QueryMaker; as a social one: new collection implementations (MD)
lacking a decent PlaylistProvider.

Even though 2.3 has gotten some excellent comments there are still
very vocal, flat out hostile rejections that are hard to ignore: there
is no decent playlists synchronization and some implementations are
flat out missing.
Not surprising as there is only one time strapped developer working on
this. Since Jeff has already reverted this I'm sure people are not
interested in improving this though.

Bart