Re: [PATCH] tracker: use "nfo" ontologies for media

[Jussi Kukkonen <jku linux intel com>]

On 02/02/2011 10:12 PM, Juan A. Suárez Romero wrote:
> On Wed, 2011-02-02 at 19:33 +0200, Jussi Kukkonen wrote:
>> nfo:Audio, nfo:Video, nfo:Image (file objects) make more sense than
>> the nmm counterparts (multimedia objects): as an example, not all
>> Photos will be marked as such, they may be just Images.
>> ---
>>
>> There is a theoretical possibility of problems here: nmm:MusicPiece is
>> _not_ a subclass of nfo:Audio (this is a tracker bug, see 
>> https://bugzilla.gnome.org/show_bug.cgi?id=641248). In reality this
>> should
>> not break anything as all tracker extractors currently set both types.
>>
> 
> 
> Actually, this patch is breaking browse hierarchy: I'm getting some
> videos when browsing through Music category.

Thanks Juan, I missed that (teaches me to test with only some types of
items in my test library...).

The ontologies are a bit weird/incomplete here: nmm:Video is a
superclass of both nfo:Video and nfo:Audio. This makes some sense on
it's own but is fairly inconsistent with the lack of nmm:Audio --
there's no way to define non-music audio (say a podcast) in the
multimedia namespace so the only way to do that is with nfo:Audio which
then gets tangled up with video...

Practical alternatives include
  A) more complex matching (grilo "audio" would mean something that is
"nfo:Audio" but is _not_ "nfo:Video").
  B) pretend all audio is music: IOW match only "nmm:MusicPiece". I'd
still continue using nfo-namespace for other media.

Option B is what most people seem to be doing at the moment -- the
downside of this is that if anyone later tries to do the right thing,
their data will appear broken in other apps (this is how I found the
problem in he first place -- a tracker miner doesn't necessarily know if
an audio item is music so I thought nfo:Audio made perfect sense in that
situation). Any opinions?

 - Jussi