#musicbrainz

so yeah, in the case of "crosby stills nash & young" the artist is a band which happens to have the same name as their members last names, so to say

so you would have to do a artist lookup on musicbrainz to get a list of the members of this band

i don't think that there is a way without the additional lookup on musicbrainz

"But my scans are dark and full of terrors"

?

Or is it horrors?

no clue as to your memetic reference

http://musicbrainz.org/ws/2/artist/46a782ea-430... there you get a list of the band members

here's another good example: Medeski Scofield Martin & Wood

I'd interpret this more as "Medeski Martin & Wood" + "Scofield"

ah yeah artist-rels

hmm

it would appear that murdos' bandcamp script is currently breaking the ability to buy anything from it

hmm, could be "MMW" plus "Scofield". I mean, a case could be made for sure.

Jazz is hard. The artists are often pretty promiscuous.

I can disambiguate a lot of that personally

but yeah

My use cases are a bit off from the norm

from time to time those collaborations are a bit finicky because it's not so clear if it's should be considered a project/collaboration or a new band

Flora Purim & Airto Moreira is another hard one

But if you're looking for an underlying logic, it's usually this: does the artist Foo&Bar have relationships that require that entity to exist? Like a webpage dedicated to Foo&Bar or something?

well, I favor decomposition, so I don't mind having two artist mbid where it's considered one because I will "and" any queries for both

I am looking for an underlying logic yes!

:D

that doesn't explain Prince & The New Power Generation though?

hysterical raisons.

yeah, I'd split that one

assigning multiple artists to a release/recording was probably not possible at first

How about Paul McCartney & Wings? :D

yeeeargh: yeah, multiple artists was a feature of "ngs", the "next generation schema"

released at the same time as the big site redesign, etc.

clearly it was added after the fact...hey my software dates back to 1998 and I am just adding that idea because of MB

(if you ever feel like a blast from the past, the old musicbrainz db can still be seen at http://classic.musicbrainz.org/ )

doesn't look that different *ducks*

if it doesn't look different than a duck, and...

cover?

word

you guys are a quack up

ugh

differing artwork for the same release

http://musicbrainz.org/release/929b4e64-af59-41... vs. https://drinkbox.bandcamp.com/album/guacamelee-...

it' all about choice, right?

anyway, rather than worry about splitting or not splitting CSN-and-sometimes-Y or MMW+S, I'd like to have a better way to build an aritst's discography based on performance names, eponymous bands etc

I suggest looking at whether an artist has recorded anything on their own

if they have, then IMO they should be split

.

er, oops,

one character at a time, like the rest of us?

akk. I'm worse than usual today.

Another issue is...given the Miles Davis Quintet ... how do we find out just the albums with John Coltrane on them?

Hey Mineo, got a question for you sorta re: an edit you made last year

_Dave_: generally, you'd have to look at performance credits on the recordings.

But in this case, wasn't Coltrane a stable member of the quintet? I can't remember.

either way

I'd like to find this information

oh, oops. no, he was only in the first quintet.

it looks like I'm going to eventually have to make a postgres slave for MB

on FreeBSD

so I can really look at all this data without impacting your main servers

Depending on what kind of queries you want to do, you could also check out https://github.com/redapple/sql2graph/tree/mast...

I haven't tried it, but presumably it's optimized for following relationships down paths.

argh!

I absolutely loathe large SQL queries like that...personal issue

thanks for the feedback