CUERipper getting weird Years from Discogs

2015-01-25 18:56:16

A bug report(?):

CUERipper 2.1.6 (Dec. 8 2014 build) is giving me random dates for the 'Year' field in Discogs-sourced metadata:

• A 1992 CD that has only ever had 1992 as its release date in Discogs came up with 1990 in the year field.
• A 2001 CD that has only ever had 2001 as its release date in Discogs came up with 1987 in the year field.

The ReleaseDate field seems to be OK.

I never noticed anything like this happening before. It seems to have just started.

CUERipper getting weird Years from Discogs

Reply #1 – 2015-01-25 19:07:30

Some more info that might be helpful.
Do you have the TOCIDs and what's your Metadata Search setting? (Default, Fast, Extensive)

CUERipper getting weird Years from Discogs

Reply #2 – 2015-01-25 19:50:47

Extensive.
iuaOow8gxVR_CQvdCrlYNYjMvuo-  1992 CD, 1990 Year
FS2SFycUwFppxBRQiPdOye5HQfY-  2001 CD, 1987 Year

CUERipper getting weird Years from Discogs

Reply #3 – 2015-01-25 20:32:52

Yup that's interesting. This looks good.
http://www.discogs.com/release/795098
http://www.discogs.com/release/419151

but I can see where the other dates could come from (perhaps from the fuzzy search?).
http://www.discogs.com/master/view/102121
http://www.discogs.com/master/view/33040

the xml data at CTDB
iuaOow8gxVR_CQvdCrlYNYjMvuo-
FS2SFycUwFppxBRQiPdOye5HQfY-

CUERipper getting weird Years from Discogs

Reply #4 – 2015-01-25 20:44:03

Year is taken from the earliest release in this group;
http://www.discogs.com/release/1859120
http://www.discogs.com/release/110141

There's a link "All Versions of this Release" on discogs pages that lets you see the whole group, or look at the 'master release'.

CTDB always did this. That's the whole point of 'year' field. Maybe those releases were recently joined to a group on discogs, or you just didn't notice it before.

CUERipper getting weird Years from Discogs

Reply #5 – 2015-01-25 21:41:28

OK, I think I did notice it before and just forgot that it was using Master Release info. It's relying on pointers from Release Groups on MusicBrainz, right?

Most of the time, the algorithm is probably guessing correctly. The Master Release will have, say, all editions of an album, and you want the Year to be when those songs came out originally, and ReleaseDate is when your particular CD came out. At least, that's what I want. I think most people are used to less robust metadata, where if the CD was made this year, that's the Year on every song, even if the songs are old. For albums it's probably right almost all the time.

With dance singles like that Den Harrow, it's not so ideal, because the remixes are from 2001, but Discogs policy (being debated right now actually, but probably not going to change) is to keep all editions of singles together, so it is lumped in with the original 1987 versions, thus the metadata search thinks 1987 is a good Year to use for the music. The 2001 version of the Den Harrow project actually has completely new vocals by some other singer, and may qualify for its own Master Release, but this 'artist' has always been like that, so who knows. Anyway there's not much that can be done about this... how is CUERipper/CUETools supposed to know if it's a remix single (new songs) in the Master Release?

The B-52s one was an error... someone saw a 1990 copyright date on the generic artwork on the promo CD and thought that was the release date. They didn't realize all of the promo CDs with that artwork have that date on them. I have blanked the date on Discogs now, so 1992 should be the earliest one it finds now.

Thanks for the explanation. Glad it's not a bug.