#metabrainz

return nan```

mayhem: not sure i understand what you mean. can you explain?

btw the normalized tables just finished, its in `spotify_cache` schema on gaga if you want to look

yvanzo: keys are still at the office

yvanzo: come to the office first

mayhem: there's 145 rows at least where spotify artist id is same but the artist name is different. i guess artist ids are not unique across artist credits but also weird that there are only 145 such cases in the data. https://www.irccloud.com/pastebin/58Td9YOk/

zas, alastairp: Can anyone bring it to me at the station?

if its correct then the number is small enough to ignore i think. also many of these would go away with unidecode and lowercasing.

yvanzo: one sec, we're organising it

lucifer: tracks in Spotify have a list of artists, right? I just want to make sure that we keep them separate and not conflated them

-d

mayhem: yes, i added a `rel_track_artist` table. which has a track_id and artist_id. so we can get the artists for each track separately.

yvanzo: aerozol is going directly to the lodging

Ah, yes, right. All good then.

if you have an example at hand where tracks from same album have different artist lists, i can check the tables to confirm

Thanks

mayhem: takes 13mins to execute.

i guess may take 30mins if cache is not warm but probably not more. we'll see in due time anyways.

Wow, nice

i'll update the branch with this query and we can try running the PR then

for dinner tonight (for those interested) we meet at the office around 19:30

so, we've just spend the last hour talking about OAuth and we have a bit of a plan for tomorrow.

are you available at 11am BCN time to join the "fun"?

yup i'll be around

the basic idea is that we'd update the OAuth lib in meb.org and hard code a couple of user accounts with a couple of scopes.

then setup a new VM with meb.org on and another VM with an lb sandbox and then test that things work.

then bring up an instance of MB and have it auth to the meb VM and slowly bring things into a prototype.

I think it might be best to include you via zoom in our discussions in the morning.

there's some code here which might be useful in updating oauth in meb.org https://github.com/metabrainz/metabrainz.org/co...

iirc its should be close but i never got around to test it.

oh great. I think the meb.org work is going to fall on you and I anyway, so that should be fine.

makes sense

once we get the VM stood up, we can start the actual work on creating a new user table in meb.org

there's a test.meb available which also runs a different database than prod so meb.org side setup should be easy.

yvanzo: are you in?

zas: yes

s/HOLY/office/

yvanzo are you at the Airbnb ?

bitmap: no, at the office

mayhem: huh there are apparently ~108,000,000 rows in track table.

Yeah I saw. Sorry I opened your bedroom door thinking it was a bathroom :)

But I'm stuck in front of her door

lucifer: yeah, I knew the number had to be very big.

i wonder what's going on with that. that's 3-4x the number of MB tracks.

I’m on my way to the office again shortly

spotify and apple music are in a quantity arms race. apple just announced that they have 100m tracks.

a vast amount of that are duplicates or listenable shit.

yeah indeed. 😞

anyways 2-3 hours it'll be done

noice!

!m lucifer

You're doing good work, lucifer!

bitmap: ok, waiting for you

[musicbrainz-server] 14reosarevok opened pull request #2675 (03master…more-flow-strict): Make more files flow strict or strict-local https://github.com/metabrainz/musicbrainz-serve...

mayhem: another things, i was thinking to rework LB#2167 as a troi patch. thoughts?

Import collections as playlist: https://github.com/metabrainz/listenbrainz-serv...

to make playlists from collections? seems a good use yes.

yup

👍

Also, everyone's tiny project is by now in Spotify and iTunes

yeah, makes sense

4 times the amount of tracks in MusicBrainz is actually a lot less than I would expect

Yvanzo: the rest of the group is supposed to meet back there at 7:30 I think?

but just 145 artists having multiple artist credits in 108M tracks is surely mind boggling.

Wait, what is that supposed to mean?

That only 145 appear in different combinations? then that cannot be right I don't think

reosarevok: https://www.irccloud.com/pastebin/58Td9YOk/

Oh, so like the same artist being mapped to two ways of writing it?

i mean same spotify id but different artist name on some tracks. iiuc i am trying to compare it to artist credits in MB

That's still weird, but less weird than what I was thinking

I didn't know that was a *thing* even

Maybe it's the result of merging artists or something?

how an artist may be credited with different names on different tracks so we have artist credit and artist credit name tables.

assuming spotify_id is analogous to artist.gid in MB then i'd expect a lot more results here.

lucifer: 11am in Barcelona we'll start our mini oauth hack day

sounds good alastairp. i'll be around

we wrote a task list, we'll share it tomorrow

great, see you tomorrow!

maybe we can put you on a video call to be along side us "in person"? :)

yes that sounds good.

I'm at the office

zas: bitmap and I are on your way :-)