#metabrainz

Hey

last week I reviewed a bunch of Picard PRs, we have planned a release this week

(Only reosarevok, CatQuest, and myself left for reviews. Last call for anyone else who wish to give review!)

apart that I investigated a warning we get about an NVME drive on trille, and decided to ask Hetzner to replace it

It will happen on Thursday, this week, 16:00 CEST so we need to ensure the server is ready for shutdown before that

👍 it's in my calendar

plus usual supervision, upgrades, user support. fin. reosarevok ?

Hi!

This week I mostly worked on some schema change upgrades, and on an official React paste a date implementation

I also saw my first wild moose, which was neat

(bring back MoosicBrainz)

Moose!

paste a date \o

paste a date \o/

This week I expect to keep concentrating on schema change stuff mostly :)

CatQuest: what about you?

Meow! `😺`

Not much this week, it went by doing misc stuff too fast

including Valborgsaften!

In that vein, Happy 1st May everyone! ♈

this week I'll be continuing doing stuff, been editing more lately

fin, go Freso !

aerozol provided a new graphic for the forum, which I added. Other than that, the usual dealing with reports, flags, being around/about.

fin.

And that wraps up reviews. Thank you all for yours. :)

catquest (jwflory): Telegram-bridge and #musicbrainz

Thanks Freso!

hi!

so, as some you might be aware, we have a telegram bridge in #musicbrainz-telegram

it was placed there instead of in #musicbrainz/#metabrainz, because there was an issue with [off]

I've learned that this issue is no longer! this feature is avail upstream

I've spoken with jwflory the maintainer of the bridge and they have said that they are happy to update the config to point to a different channel. They've also expressed the need to voice the tg-bridge bot and such, to prevent it getting banned.

https://chatlogs.metabrainz.org/libera/musicbra... for log

of discussion

so iwas wondering, what needs to be done/discuss how to enable this for us?

Freso: what do you feel about that?

I can do the +voice for it. I don’t recall any other objections to it, so it’d be fine for me, at least as a test at first, and if no one complains, we can have it be more permanent.

You know IRC better than most

+1 for me in that case

does it do any stupid stuff like matrix when writing messages longer than the IRC limit, or posting images etc?

If no one else have any objections, let’s go ahead with that.

well...

how many active people are in the telegram channel that are unable to use something like matrix?

(I note that there have been about 3 things said in the channel in 2022)

atj: Not that I can recall, but join #musicbrainz-telegram and try it out?

so it seems it's not gonig ot be too annoying for users in mb i guess?

I think it might be more active if we actually join it to channels we actually use :)

my thoguhts exactly

alastairp: I think part of the lack of activity is also people just being in the (tg) channel but not writing, which might be more active once they get the traffic from #musicbrainz :)

yeah, perhaps

So happy to test it and keep it unless it does annoying stuff

Might be a chicken and egg thing - more might use it if hooked in?

can we have a probation period of 4 weeks or something?

Which is why I say we go for a "live test" run and see what happens.

and then discuss it again

not a bad idea

I expect disconnecting it if it's annoying is just unvoicing a user?

yea

If so, 4 weeks sounds good to me

reosarevok: Well, at worst case kickbanning it, but yeah.

The voicing is more preventing the network from kicking it off, not from keeping it out of the channel.

yes

Seems like we have a decision

(And tools like ozone for example are a lot more lenient with voiced users than non-voiced ones too.)

Freso, maybe you can help making it happen then? :) No huge rush I guess

yea

deff not a rush

Yeah, I’ll talk with jwf about it.

:D

fin?

nothing more from me

I think that’s fin anyway, so…

\o/

lucifer, zas, et al: Securing MeB infrastructure - part 3

yvanzo is not around, should we keep the security stuff for a day he is?

yes

yup makes sense to do it another day

Could be we just arrange for another security meeting tomorrow at regular meeting time…… :)

(If yvanzo can make it then.)

yvanzo is supposed to be around tomorrow

Looking at schema change stuff, but maybe we have time for both? :)

So we could schedule and if he asks to postpone we do

Or if nothing’s urgent, we can just push it to next week?

Seems like that’s less moving pieces.

next week sounds good then.

let's just go with next week rather than splitting yvanzo's time tomorrow

Ok :)

next week is getting even closer to schema release - do we see anything urgent on the doc that should be discussed before the 15th?

(unless MB team are happy to do it next week)

I'd say if it fits in next week's meeting, let's talk, if not let's decide next week when to do it

(probably after SC)

only incident response section is left.

So, let's see how long next meeting is? :)

+1

Ugh. _Now_ the delivery person came.

Anyway.

Seems like we’ve agreed to postpone to next meeting, so let’s do that.

uh, btw - did we miss bitmap from reviews?

no, he's not around this week

ah, cool

Which then concludes this meeting. Thank you all for your time. Stay safe out there and wear your masks! :)

</BANG>

I am that freak who continues to wear masks in confined spaces

try telling that to most people here :/

mayhem: got a moment to talk about cache tables/batch updater?

mayhem: same

but: mask

mus*

NZ is good about wearing masks still

fff

bus: mask

mayhem: Same. (Also in non-confined spaces, since I don’t want to fiddle with taking it on and off.)

aerozol: you live in a reasonably sane country

alastairp: yep

mayhem: https://gist.github.com/alastair/278aa750266613...

Mayhem, lucifer: had two thoughts! Wonder if there's appetite for a case study on your road to recommendations/playlists once your done? As a blog post. By digging through chat logs and/or taking to you

what would a case study do here?

this log format threw me a bit - it's not saying that 100k/104k = 1.4%, right? instead it's saying that it processed 104k rows, turn that into 100k inserts, and that's 1.4% of the 7.7m rows that it selected?

The other thought, much more fun: Can we please have visualizers for listenbrainz haha. Bring back skins!!

mayhem sure has been working on that for far longer so he can assist better then but happy to help as much as i can :)

I think a lot of people would be interested in reading it

alastairp: yeah, I suppose that the # of inserted rows could be moved to be the last column and that would more clear, I think.

Are people aware that MB is doing stuff like this? It seems like you're drawing on and maybe creating some pretty leading edge stuff

visualizers like the windows media player ones?

Yes!!

however, we are doing a few things that no one else is doing -- transparency in recommendations.

"processed 104,861 rows (100,000 inserted). 1.4% complete"

oh that could be a nice thing.

This isn't a serious suggestion... unless...

;)

one of the things that I want to work on is giving user more control and giving users an idea as to why we recommended a track.

that might be "leading edge" -- the rest is kinda bog standard.

Even the philosophy of why it's important... Super interesting

the tricky part is that we're not quite there yet.

mayhem: next: https://github.com/metabrainz/listenbrainz-serv... the behaviour here is that if lb_conn is set, it inserts into that db handle, otherwise it inserts into mb_conn

we will be soon.

I've watched you both do some things that didn't work and then some little things that did - really interested in that stuff

what does a case study attempt to do/solve?

This is my marketing/comms hat

If awareness of MB helps you, then this helps. For users or orgs

If not, then it doesn't :p

aerozol: well, a lot of times it is trowing things are the wall to see what sticks. lucifer and I are understanding more as we go, but this is somewhat opaque stuff we're working on.

the mb_metadata_cache inserter requires that there is a canonical_release_redirect table in the musicbrainz database to join against, so is there a recommended way to make this happen other than manually editing config.py to remove the timescale connect string temporarily while we run this, and then adding it back?

alastairp: that is the only mechanism for controlling the destination of the tables right now.

Maybe 'blog post' is less annoying sounding than 'case study' and I should have said that :D

ok

it would be nice to have an extra layer where config.py just specified valid connection strings and some config that explicitly directs tables to be written in one of two (or both!) places.