#metabrainz

yvanzo: no, and unfortunately, neither did installing setuptools==65. same error in either case: setuptools.extern.packaging.version.InvalidVersion: Invalid version: '2.20201112.1-7-g711fb3b'

this is in an Python 3.11.6 virtualenv

tried combining them too https://gist.github.com/mwiencek/3df067ad5c445a...

yvanzo: higher priority because of the translations :)

rdswift, yvanzo, aerozol: nothing we were talking about today is connected with wiki pages anyway :)

rdswift: so please don't be discouraged, I'd still love to move the stuff away

These "docs" are actually just existing admin interface for editing stuff, just made visible for users, not connected to the usual docs system (like relationships pages are)

aerozol: guidelines can (and IMO should) also be moved anyway - there's only stuff like "I, a user, wrote this quick guide" that IMO should stay in the wiki, plus any stuff that actually needs to be a wiki, like i18n pages that need input by translators

[musicbrainz-server] 14reosarevok merged pull request #2730 (03master…more-perl-critic-modules): Follow Perl::Critic::Policy::Modules rules https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] 14reosarevok merged pull request #3094 (03master…MBS-13368): MBS-13368: Properly compare series relationships if series is target https://github.com/metabrainz/musicbrainz-serve...

yvanzo: do you remember if we have any way to know how many pieces of cover art a release has using the MB API before calling the CAA one?

I *think* there is not, but maybe I just don't know about it

(someone was asking in support about finding the release with the most images for a RG)

yvanzo: I guess https://github.com/metabrainz/musicbrainz-serve... is fine to merge with that hack then. If you do I can rebase my lowercasing PR on it

rdswift: reosarevok: if we do a shift them I don't know about having guidelines in the same docs tbh. Or maybe we can make the colors distinct for each section or something

Seems like it might present an opportunity to make the different types of 'doc'/'help' etc clearer

Not just because I think it's a bit confusing, the survey results I'm (still) working on show a lot of users are confused with the docs as well. It's in the top 10 'what's your biggest issue with MB' ranking (#10 but it counts!!)

Why do docs need to have a theme though 😂

A shift is better than nothing but I'd like to brain nstorm if we can get any extra, preferably easy, wins while we're at it

Just made it doc doc stuff

Make*

akshaaatt: what are you replying to sorry?

Ah, I was reading through previous chats and it said to have a theme for docs as well

akshaaatt: because read the docs by default is hideous and nobody would want to look at it unless they're a dev :p

Makes sense reosarevok

Ah right - didn't you tell me at the summit you hated the look of ReadTheDocs :p

(and our current user docs look like pages inside MB proper, so the closer we can get to that, the less of a clash it will be)

Ik of a good paid tool called gitbook which is great for guidelines and product documentation

An interface like that would be cool

Don’t know of relevant oss solutions though

aerozol: yus! For guidelines stuff, rtd looks bad

My more recent message re. colors is to theme stuff to differentiate between the very different types of doc that MB has - dev documentation and editing guidelines are completely different things imo. If that helps explain it a bit

Yus agreed

O’Moin

bitmap, lucifer: I found the core issue with mb-rngpy and pushed a tag as a workaround for testing sir in the meantime.

reosarevok: I’m just encouraging rdswift for now to not waste too much his time which I value.

Sure, I just don't think it will ever be a waste :)

There's no need to work super hard on it now but if aerozol wants to help, why not - it's not like either of them are going to update the search server:)

Just to clarify, the documentation has been among my top concerns for years, I’m not underestimating it.

yvanzo: makes sense, thanks

mayhem, monkey, aerozol: here is what the tag/genre coverage in MB looks like at the moment: https://usercontent.irccloud-cdn.com/file/n0eFm...

for the genres in YIM, how do we these?

if a genre is present on the recording use that, else try release, otherwise try release-group and finally artist?

or do we use all of the entities and assign different weights to genres based on what entity they are coming from.

in LB radio I just flatten all the tag data for a given entity and use total counts and that works ok. I suspect you can do the same here.

at least you could start with that and see how the data turns out.

mayhem: not sure what you mean.

we want to find the top 5 genres a person listened to in a year here.

if a user listened to a recording that has an artist tag of pop and a release-group tag of pop, then total that is two counts for pop.

i see makes sense

👍

mayhem: for lb radio, we don't use release tags. same for yim?

dont we conflate release and release-group tags?

I'd say use all tags you can get your hands on.

oh right we do. i had forgotten about that 🤦

sigh. spotify lays off 17% of its workforce and the goal of getting paid before year becomes a fantasy.

It's kind of funny how that seems like an insane amount, but if MeB was laying off 17% I guess it'd be... one person? Two? :D

But yeah, everyone seems to be struggling and/or trying to cut costs for profit by looking like they are struggling recently

Well, hopefully they will pay *ever*... :p

more the latter than the former, reosarevok .

my guess: VC: you're too heavy, no more money until you cut headcount.

and out comes the axe

*schwiiing*

lucifer: I keep being stuck with this build problem on listenbrainz. its not related to any branch. I've removed all images and all volumes and the problem persists.

doesn't happen on any other machine.

I guess I can try to uninstall docker and re-install.

but I'm unable to work right now. :(

any ideas at all? I'm stumped.

mayhem: hmm weird

try a no-cache build locally

I deleted ALL images and ALL volumes.

there should not be anything to pull from cache. and that no cache option was the very first thing I tried. :(

`docker system prune -a` maybe?

`docker system prune -a --volumes`

to remove every bit of docker data.

I can try that next time, just uninstalled docker, decalring total bankruptcy.

that. :)

which os are you using?

ubuntu (pop os)

it was time to upgrade docker in any case. rebuilding now.

lucifer: that fixed it finally.

awesome great

i would even suggest to start using docker compose v2

its faster, at least feels so

is that docker compose vs docker-compose ?

yes

yeah, been there a while

oh interesting

did you alias docker-compose to docker compose? because every now then i keep encountering a meb project with develop/test script that barfs on docker compose

nope

i guess meb.org and lb were fixed sometime back so rarer for you to encounter it

[listenbrainz-server] 14amCap1712 merged pull request #2651 (03master…fix-playing-now-cover): LB-1371: "Playing now" does not show cover art on page load https://github.com/metabrainz/listenbrainz-serv...

yvanzo: ping :)

ouch ;)

why you gotta ping so hard reosarevok ?

yvanzo: if you merge the admin thing and I rebase the other string PR, could we still release beta tonight and freeze, what do you think? :)

I'll try to ping softer next :D

monkey: after all those changes, I *finally* got the branch up. lolsob

I also merged master into the branch.

Woot

[musicbrainz-server] 14yvanzo merged pull request #3091 (03master…mbs-13117-l-admin): MBS-13117: Prevent localizing admin-only messages https://github.com/metabrainz/musicbrainz-serve...

reosarevok ^ done

[listenbrainz-server] 14amCap1712 merged pull request #2652 (03master…fix-follow): LB-1315: Attempting to follow a user while not logged in gives error https://github.com/metabrainz/listenbrainz-serv...

mayhem: https://github.com/metabrainz/listenbrainz-serv... and https://github.com/metabrainz/listenbrainz-serv... , let me know if you'll get time to review these today. if so i'll wait otherwise do the release now.

also is https://github.com/metabrainz/listenbrainz-serv... ready for review?

its marked such but tests seem to be failing

yeah, sorry the test are still failing and I have zero idea why. if you have a clue, let me know.

as for the others, trying to work with monkey to day, but let me look real quick.

I'll get to them in the next 2 hours, how is that?

sure no hurry

i'll take a look at the tests

thx

you are satisfied with the similarity scaling?

yeah, nothing ever changed. I think the value as we receive it is already a percent and we should simply use it.

and the confusion of having two columns in that table is also a pita.

hmm so do we even need to merge that PR?

it would be good to get rid of the second column, we no longer need it.

i can fix that two column thing to separate columns.

yes, we still need the PR.

back and frontend changes that make the presentation to the user better.

we were using the global similarity to find spammers

we should just have one column and be done with it. but then we need to be more careful about how to release this thing.

another thing you might help me with real quick, lucifer :

in this query I want to fetch the release_group.artist_credit and explode that into a list of artist_mbids. I always get that query wrong and you always have to fix it. what is the correct query fragment to do that?