Notes from #MetaBrainz Meeting 2023-07-03

[aerozol]

MetaBrainz Meeting 2023-07-03

Meeting start: IRC Logs for #metabrainz | MetaBrainz Chatlogs
Meeting wiki page: MetaBrainz Meeting - MusicBrainz Wiki

Agenda

• Reviews
• reCAPTCHA
• Google Fonts
• LLM policy

Last meeting: 2023-06-26
Next meeting: 2023-07-10

@reosarevok: Welcome to “Half the year is gone, how come” Monday!

Reviews

@atj

[mailed in] Last week I continued working on the openresty and consul-template ansible roles and migrating the complex configuration.
I also finished implementing a new role for managing apt package installation which will make life easier now we have some servers running Ubuntu 22.04 as well as 20.04

@aerozol

[mailed in] Getting packed for my holiday! I’ll keep writing meeting notes (which lets me keep up with MeB happenings as well), but not within the usual 24 hours.
Last week I did some fun stickers with mayhem, and worked on some long overdue LB release page mockups.

@yvanzo

Last week I looked into GDPR vs. reCAPTCHA & alternatives for MBS-13146 [Registration form may not conform to the GDPR].
Also pursued setting up and documenting the translation platform.
Plus usual support and al

@reosarevok

Last week I didn’t do a lot since I was kinda resting, but I still dealt with a bunch of reports, helped yvanzo with documentation for i18n
And looked into MBS-13139 [Stay on same page when going to main server from beta site banner] (but I think that still needs a bit of work)
This week I’ve so far worked on a bunch of small guess case issues

@snaekboi

I worked on fixing my local docker setup, which went smoothly, and finally got things working as they should
Now I’ve been writing some tests for the new features!
Also with the help of bitmap, I closed PR #1 and opened two new PRs for the musicbrainz-bot

@bitmap

slow week for me, but mostly similar to the previous: some code review, gsoc, fixing issues with the pgbouncer container (still need to test it on hetzner servers), and testing a few new Flow features in the MBS codebase

@CatQuest

Did some instrument images because the weather has ben shait, it’ll be again this weekend so more instruments to follow♊

@kellnerd

I’m through another week of debugging the OpenLibrary importer’s producer script, which was hanging instead of running or at least throwing an error.
The script basically has two asynchronous parts where I suspected the error: 1) Streaming lines from the database dump, 2) Library to push messages into RabbitMQ
Both worked in my separate testing with minimal examples, but nevertheless the idea seemed plausible as there were reports online of similar problems for each of them.
But after having wasted multiple days with debugging this (commenting out various combinations of lines), I found the real culprit:
It was the import statement for the logging library, which caused side effects which were severe enough to get the RabbitMQ library hanging
I kind of guessed it would 've been faster if I had rewritten the script from scratch and I was right.
I probably wouldn’t have used that library.
Weird enough nevertheless. Now I still have to fix some things, but I hope I’ve dealt with the worst.
Other than that I reviewed a few userscript PRs.

@mr_monkey

Last week I worked on a tag voting/creating component for ListenBrainz, happy to say today it is finished: https://github.com/metabrainz/listenbrainz-server/pull/2475
Currently deployed at Sign in - ListenBrainz for those of you following at home [note: tags will be submitted to the live MB database]
Also continued refining the new “created for you” page which is also pretty much finished: Redesign Recommendations page by MonkeyDo · Pull Request #2503 · metabrainz/listenbrainz-server · GitHub
I reviewed some PRs and helped GSoC projects along.
Also had a good video chat with aerozol, reviewing the design of the above PRs and planning for the next steps (LB album and artist pages).
And finally did some general housekeeping and ticket triage

@lucifer

last week i worked on the migrating the tag endpoints to spark and LB db so that mayhem could use those in tags-radio.
then i worked on adding support for exporting batch generated weekly playlists to spotify.
also reviewed a couple of PRs.

@zas

last week I spent some time on investigating an annoying issue concerning Picard threading and remote command pipes, a patch was made by phw today based on my findings
I also improved monitoring regarding a keydb issue we have on rex/rudi, and updated informations in upstream bug report
apart that, I did usual upgrades, user support, supervision and minor fixes.

@rob

last week I mostly worked on LB radio and getting a few new features in.
I’ve improved the syntax and made it consistent with MB – hopefully in the next day or so I can integrate lucifer’s new tag data, which brings the lb-radio to a solid alpha phase where I’ll ask for feedback from the community.
I’ve already found and rediscovered music, so I am very excited by this!
I did a few other projects here and there, small stuff.
reviewed Arsh’s work, which is starting to look really nice.
and today I descended into accounting hell working out the finances from the last quarter and paying people.
I’m quite curious to get our half way through the year report card and see how it stacks up to the plans made in january.

@arsh331

Last week I continued progress on my GSOC project for artist similarity graph
Made small adjustments to achieve the right balance of things
Most of time went into fixing the scaling function for the graph and setting up the colors for the same
Overall I am happy to learn many new things

@jasjeet

just finished phase 1 of my gsoc project yesterday
trying to brush up any remaining edges or any better ways to route state
also practicing my xml skills again

@akshaaatt

I did the usual PR reviews and thought about a few things
Other than that, my head has been spinning for the last week or so with the amount of continuous learning and management of things.

@Vishalftw

I am inching closer towards trying to make progress before the mid term evaluation
I tested the implemented query changes in data set hostler repo this week i.e Converting models in queries to pydantic type. With that I also migrated one of the api query in listen brainz repo, most of the api’s left in it which should follow the same pattern.

@vardansaini

I continued working on Apple Music integration
I have fixed the preview playing issue and addressed to most of the comments on PR
facing some issues with refresh token
as apple does not provide them
but I was looking at the chat and saw the doc by Lucifer
hopefully we will be able to overcome that
still working on error handelling and stuff

@ShivamAwasthi

last week I worked on creating middleware for privilege-based access to routes
I also started writing tests for the same, but wasn’t able to make as much progress as I wanted to because I went on a mini-trip with family.
This week, I aim to finish writing tests for these routes, as well as complete the work on Admin Logs page.

@riksucks

reviewed vscode_'s PR and working closely with him for shipping it

reCAPTCHA

(https://tickets.metabrainz.org/browse/MBS-13146)

@yvanzo

Our current use of reCAPTCHA in MB has been questioned with respect to the GDPR.
See MBS-13146 [Registration form may not conform to the GDPR] for details.
Basically alternatives are cookie-free captchas, honey pot, nonce, and form delay.
These can be used together too.
I don’t really have any question now I made some research about those.
Is it alright to move away from reCAPTCHA?
Extra question: Should we move away from captcha at the same time?

Read the chat logs for the full discussion.
It is agreed to move away from reCAPTCHA, yvanzo to carry on with finding a replacement.

Google Fonts

(https://tickets.metabrainz.org/browse/ORG-51)

@rob

Google Fonts is basically the same sort of question: Google hosted Google Fonts are basically a GDPR violation.
so, can anyone see a reason why we should not self host the google fonts our projects use?

Read the chat logs for the full discussion.
It is agreed to self host. Everyone gets a ticket.

LLM policy

[continued from the last meeting]

@aerozol

[via reosarevok] The community feedback is all collected in this forum thread.
UltimateRiff posted a very useful poll, which had 22 responses: 63%: A, Allow no LLM, 27%: B, Allow some LLM, 9%: C, Allow all LLM
There was some interesting discussion, but the poll should be enough to deduce the feeling of the community.
The community shared no substantial feedback re. whether we address this in the MeB Code of Conduct, or on a per-project basis (one comment in favour of MeB-wide/CoC).
The goals of this (final?) LLM meeting topic should be to pick one of the options (A, B or C), and to raise any comments that you want taken into account re. guideline wording.
I (aerozol) or someone else/a group can then go off and create a decent guideline/code taking that feedback into account, too avoid bogging down this meeting.

Read the chat logs for the full discussion.
Mention was made of the Wikipedia draft LLM guidelines that @derat shared.
“general consensus is that we dont want LLM shit, but its not a 100% hardline.”
e.g. a fair amount of slight disagreement from meeting attendees, but nobody voicing that we should allow all LLM.

Conclusion: Hide these and new posts and then when aerozol comes back, start work on the policy.

[UltimateRiff]

@mr_monkey

Last week I worked on a tag voting/creating component for ListenBrainz, happy to say today it is finished: Listening now: allow users to vote on and add tags by MonkeyDo · Pull Request #2475 · metabrainz/listenbrainz-server · GitHub
Currently deployed at Sign in - ListenBrainz for those of you following at home

looking forward to this going properly live~

something to note for those of us following at home, even though it’s on test.listenbrainz.org, it does actually submit the tags to MusicBrainz. I didn’t expect this, but that’s probably because I’m used to seeing test.musicbrainz.org which doesn’t use the live MusicBrainz database. might be useful to note this in the meeting notes even