#metabrainz

[musicbrainz-server] 14derat opened pull request #3510 (03master…info_bubbles): MBS-5944: Rework doc bubbles for release editor info tab https://github.com/metabrainz/musicbrainz-serve...

lucifer: ping me when you're around

[@zas666:matrix.org](https://matrix.to/#/@zas666:matrix.org) hi, yes will be around in a couple of hours

[metabrainz.org] release 03v-2025-04-02.0 has been published by 14mayhem: https://github.com/metabrainz/metabrainz.org/re...

zas: i think i won't be needed the vm after all. i figured out a solution between asking for the vm and now. i'll let know if things change again 😅

mayhem: first draft of similarity data is available for testing.

any MBIDs you have in mind?

for musicbrainz.org/recording/60444aa3-f190-4400-a7...: https://musicbrainz.org/recording/f3bba4cd-8018..., https://musicbrainz.org/recording/8e74dd9d-e5a3..., https://musicbrainz.org/recording/1bfc064e-fcd0...

* for musicbrainz.org/recording/e97f805a-ab48-4c52-85...: https://musicbrainz.org/recording/f3bba4cd-8018..., https://musicbrainz.org/recording/8e74dd9d-e5a3..., https://musicbrainz.org/recording/1bfc064e-fcd0...

<lucifer[m]> "for musicbrainz.org/recording/60..."; <- it this MLHD similarity data?

those similarities are pretty good, but is there any way for me to check them in DSH?

mayhem: not at the moment.

yes MLHD similarity data.

ok, let me think up a few examples.

i am pretty sure there is some bug in the data because i got 42M similarity pairs which is more than the 14M we get from LB data but still not as much as I'd expect.

parameters used: --session 300 --contribution 5 --threshold 20 --limit 100 --skip 30

it may not be the pair count to judge this -- perhaps summing the counts to see if all tracks are accounted for?

can you use session_based_days_7500_session_300_contribution_5_threshold_10_limit_100_filter_True_skip_30 ?

because that is the dataset we have right now, so that would be a good comparison

sure

i need to make some changes to the code but we can run this entirely on spark cluster in ~10-12 hours.

so we can run experiments easily with this data.

possibly even ~6 hours but i'll have to run some tests.

taking it one chunk at a time, using zstd compression, changing mbids to ids, breaking the data generation into two stages.

helped ensure that we are able to process the data on the existing cluster. and its crazy because the final data (ids not mbids) is just a 180 MB parquet atm.

oh wow. you calculated this without the big new VM?

yes

amazing. this is so much better for being able to iterate this data

yup indeed

i had spent a week on figuring out to make this work lol but just after i asked for the vm yesterday, i figured out how to fix the issue..

[listenbrainz-server] 14MonkeyDo merged pull request #3241 (03master…hide-apple-music-export): Disable "Export to Apple Music" option if user not signed in, or into Apple Music https://github.com/metabrainz/listenbrainz-serv...

When do you expect the new country code from MBS-12170 to be available in the json output on musicbrainz.org (or beta.musicbrainz.org)? I'm looking into adding this into the variables available in Picard, and I want to make sure I'm extracting and testing properly. Thanks.

MBS-12170: Add artist country in release webservice calls https://tickets.metabrainz.org/browse/MBS-12170

augh, co-opting the artist *country* because releases have useless information is s...

now adding it to the api I'm not against, but the real issue here is that we need to stop/revert/codify to prevent these "205 country" monster releases

sigh. it was literally what the [worldwide] thing was supposed ot be *for*

sigh

<BobSwift[m]> "When do you expect the new ..." <- I believe reosarevok is planning to update the beta server tomorrow

<BobSwift[m]> "When do you expect the new ..." <- I've deployed this on the test server in the meantime