duplicates plugin thinks there is a duplicated recording in an album with actually a duplicated recording.

[doronbehar]

Problem

There are releases in the musicbrainz database which actually have a certain recording multiple times and this causes the duplicates plugin report these recordings are as duplicated. I'll examine with the following example:
This release by Frank Sinatra: Three Original Hit Albums, has the recording of the track I Get Along Without You Very Well twice, as you can see in this and this tracks.

IMO, the duplicates plugin should check before reporting whether or not these tracks appear in the database twice as well as within the local files.

[sampsyo]

Interesting point! I'd be interested to hear more thoughts on exactly how this should work. The current algorithm for the duplicates plugin is pretty simple: look at all your tracks (or all the tracks that match the given query) and look for duplicates among them. There are no queries to the MusicBrainz Web service. There is also no interaction with the album structures in the beets database; items are processed completely independently.

How should an algorithm work that respects the fact that some duplicates might be intentional? It's not completely clear to me how this would work—or how we could make it efficient, without querying the MBWS for every album.

[doronbehar]

Well I think that even if we only check the track tag for each of those files, and see if it's different, we should be OK.

[sampsyo]

Hmm… if that's enough, may I recommend just adding track to the keys section of your duplicates config? http://docs.beets.io/en/v1.4.5/plugins/duplicates.html#configuration

That should consider items with different track values to be different.

[doronbehar]

OK.., I had no idea I can configure all of these options, it's great, but still there is one question that should bother us:

We can see that the different tracks which are reported as duplicates are:

https://musicbrainz.org/track/a3b6d068-0c45-32c1-afe7-dcedf24407fa
https://musicbrainz.org/track/f2f23b35-6928-3c6c-9c48-4b2f7ff5cf38

Both have a different ID. Are those supposed to be the values of mb_trackid? Because if so, then the problem is not with the duplicates plugin but with the value mb_trackid gets in general. It's a fact that on my machine, the value of this tag is not different for the reported duplicated files.

[sampsyo]

Yes. For historical reasons, mb_trackid actually holds the recording ID. See also #406.

[doronbehar]

Hmm, it seems no one responded @mineo's #406 (comment) which asserts pretty much what I was saying as well. I think a PR should be made.

[sampsyo]

Indeed! Any help would be appreciated.

[doronbehar]

Is mb_recordingid being used in beets' database?

[sampsyo]

Nope, we don't have a field by that name. The recording ID goes in mb_trackid. (This is because, prior to MusicBrainz "NGS," that ID was called the "track ID," and changing the name now would break some configurations.)

[stale]

Is this still relevant? If so, what is blocking it? Is there anything you can do to help move it forward?

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.