Updating:
http://musicbrainz.org/release/1e8f0eda-b120-4495-aec6-416d83820697
The problem is tracks 7-12 are the same recording ID which causes two problems:
This of course is because musicbrainz has removed the uniqueness of the track entities. I imagine this will cause problems for a lot of systems.
Hmm, this is a pain I’ll try and improve Jaikoz to deal with this but as you say it would be useful if there was a unique id for a track on a release, see http://tickets.musicbrainz.org/browse/MBS-1839
I added my 2 cents to MBS-1839.
Okay, thanks, in the meantime Ive modified Jaikoz to workround this issue ready for next release.
Fixed in 4.1.0 NGS Beta
Jaikoz is now allowing the duplicate recordings to have the correct track number but the “Delete Duplicates” isn’t working.
I’m using the setting “It has the same MB Id and Release Id as another song” which only prevents the removal of the same recording appearing on multiple releases, not the duplication of recordings on the same release.
Since the release ID applies to the whole multidisc tracklist you would have to also add the qualifiers for disc number and track number.
The summary is that the recording ID is actually irrelevant in finding file duplicates, a unique key of release ID + disc number + track number is needed.
You are right for the edge case of having the same recording on the same release more than once then running delete duplicates would incorrectly delete one of the songs.
But at least you still have a copy of the song, so you can easily make a copy of it.
Having the exact same recording on an album more than once must be a pretty rare occurrence but I’ll try and devise a workaround for it. However dropping recording id in the match and just using trackno and discno is not a good idea. The recording Id is usually correct, the Track No and DiscoNo may not be it would happen all to often that two different tracks on a release had incorrectly been edited so they had the same trackno. Matching on recording id, trackno, and discno prevents deletion of diuplicates where they data isnt quite there.
One solution I could do would be after creating a list of duplicates would be for Jaikoz to check the releases for duplicates tracks, and unmark tracks for deleltion when the same track appears twice. But this would need to send queries to Musicbrainz and hence be slower.
[quote=paultaylor]You are right for the edge case of having the same recording on the same release more than once then running delete duplicates would incorrectly delete one of the songs.
But at least you still have a copy of the song, so you can easily make a copy of it.
Having the exact same recording on an album more than once must be a pretty rare occurrence but I’ll try and devise a workaround for it. However dropping recording id in the match and just using trackno and discno is not a good idea. The recording Id is usually correct, the Track No and DiscoNo may not be it would happen all to often that two different tracks on a release had incorrectly been edited so they had the same trackno. Matching on recording id, trackno, and discno prevents deletion of diuplicates where they data isnt quite there.
One solution I could do would be after creating a list of duplicates would be for Jaikoz to check the releases for duplicates tracks, and unmark tracks for deleltion when the same track appears twice. But this would need to send queries to Musicbrainz and hence be slower.
[/quote]
It all comes back to the fact that musicbrainz has removed the track ID. I understand that reading the disc and track numbers isn’t a great solution since these tags don’t always exist in files but theres no other way to uniquely identify tracks without incorporating these values.
I would be happy if there was another option on Delete Duplicates called “It has the same Release Id, disc number and track number as another song”
Out of the 928015 tracklists, 130 contain duplicate recordings. Rare but annoying.
Actually I think the best solution would be for me to write a composite key as another field into the file when a match is based on Musicbrainz
i.e MBTRACKONRELEASEID=recordingid-releaseid-discno-trackno
I would then use this field when looking for delete duplicates in your scenario.
It would also make it easier to update the song with the correct trackinfo because I can then just use the MBTRACKONRELEASEID field to match and it wouldnt matter if the discno or track had been removed.
But this change is going to take a little bit of time.
[quote=paultaylor]Actually I think the best solution would be for me to write a composite key as another field into the file when a match is based on Musicbrainz
i.e MBTRACKONRELEASEID=recordingid-releaseid-discno-trackno
I would then use this field when looking for delete duplicates in your scenario.
It would also make it easier to update the song with the correct trackinfo because I can then just use the MBTRACKONRELEASEID field to match and it wouldnt matter if the discno or track had been removed.
But this change is going to take a little bit of time.[/quote]
Yep, no problem.