#metabrainz

lol, every time I click between ‘loved’

and ‘hated’, it adds this Hex Waves song again

(refresh removes all except the original)

If anyone wants to try reproduce: https://listenbrainz.org/user/aerozol/feedback/

Do we have all the summit recap slideshows stored anywhere?

^ spotted in the summit meetings that we were going to ask the community for feedback on this, so made the thread. Edits to the topic welcome

Also: “We agreed to have a communal security review the first week of each month. Someone needs to remember to add this to the meeting agenda at the right time.” Did anyone remember hah

akshaaatt: i'll be in delhi from oct 24-29 as well.

What do people prefer for all the MBz subsites/offerings? Products? Projects?

I’m leaning towards ‘Projects’. If we can decide I’ll put it in the writing style-guide so nobody has to ask again

mayhem: whoever else ‘does’ the blog: I spent much too long drafting a lengthy blog post for the summit. It needs an eye cast over it to make sure I didn’t make it too silly, and also I left the recap section for now in case we have all the slideshows somewhere to link to? Will write that part tomorrow either way

Dive back into how Last.fm used to look - LB’s looking pretty flash already! https://www.reddit.com/r/lastfm/comments/y85wo2...

New Last.fm feature: You can now search your library (pro + desktop users only): https://www.reddit.com/r/lastfm/comments/y80d4m...

(I never used reddit before checking on MBz stuff, if anyone’s wondering…)

aerozol: fixed. https://test.listenbrainz.org/user/aerozol/feed...

[listenbrainz-server] 14amCap1712 opened pull request #2212 (03master…feedback-page-fix): Feedback Page duplicate tracks fix https://github.com/metabrainz/listenbrainz-serv...

lucifer: I am not worthy!

hehe lol. can you confirm the issue is fixed for you?

Fixed!

👍

for MBz subsites, i too use projects mostly.

[design-system] 14Aerozol opened pull request #90 (03master…master): Adding common term for sub-brands: Projects https://github.com/metabrainz/design-system/pul...

[design-system] 14Aerozol merged pull request #90 (03master…master): Adding common term for sub-brands: Projects https://github.com/metabrainz/design-system/pul...

Thanks lucifer, I’ve put in ‘Projects’ unless/until someone says otherwise

Hey is it possible to manuallly add listens to listenbrainz?

I'd like to add a whole tracklist that I listened as a DJ set on youtube manually to listenbrainz

Pratha-Fish: you can use listenbrainz api: https://listenbrainz.readthedocs.io/en/latest/u...

lucifer: Thanks

lucifer: for last.fm, I used to use web.finale.app to lookup songs using spotify search and instantly scribble them with a click

Pratha-Fish: i guess we could. but LB already does some mapping internally so not sure how much useful it would be. if the main thing you want to do is submit listens from websites like youtube etc, you could try using webscobbler extension?

akshaaatt: can you email me the receipt for the extra luggage we added for your return flight?

i think the one issue could be that you couldn't submit things listened in the past.

🤦 I was supposed to press Shift+Enter instead of Ctrl+Enter

:D

lucifer: Pratha-Fish: user Flaky on Discord was working on something to submit listens, but says they never finished

They also pointed to: https://community.metabrainz.org/t/dealing-with...

*never finished but might get back to it soon

monkey: aerozol: sticky filters on mobile. thoughts? https://usercontent.irccloud-cdn.com/file/83UYP...

lucifer: yes exactly. I want to be able to manually enter the artist and recording name and add the listen instantly, even if it isn't being played live on my device

aerozol: thanks, I"ll check it out!

Pratha-Fish: yes, sounds good to have a UI for that. but whether that should be a part of LB or an external app/site not sure. probably best to ask for other's opinion.

[data-set-hoster] 14amCap1712 opened pull request #7 (03master…arbitrary-outputs): Add support for more flexible outputs https://github.com/metabrainz/data-set-hoster/p...

[listenbrainz-server] 14amCap1712 opened pull request #2213 (03master…arbitrary-outputs): Add listenbrainz sessions viewer query https://github.com/metabrainz/listenbrainz-serv...

alastairp, i needed more flexible output format for data-set-hoster so i propose the above changes. this however removes the summary field, the new format supports summary field in a different way. however i could not find existing uses of that field in LB. do you remember why that was added? (need to update it to the new format to ensure it keeps working)

a new wave of people on twitter is realizing that cloud is expensive.

lucifer: Sounds good 👍

thanks lucifer, looking now

Pratha-Fish: hi, how's that post going?

aerozol: I added you as editor to the summit notes

alastairp: Hi, the blog is going fine! I think I am halfway through it, and I am aiming to get it completed in the next 4 hours to a stage where we can go through it together once and make edits

excellent

alastairp: Here's what I've written so far. Feel free to suggest edits :D

thanks alastairp!

Pratha-Fish: I think it makes sense for you to get an initial draft done first so that I don't step on your toes. let me know when you're happy with the first version

alastairp: sounds great! I'll let you know ASAP

lucifer: in general, I think that it's probably fine to change this output to suit our needs whenever necessary? we make the web interface available, but don't publicise it a lot, or make any guarantees about the api, right?

but I know that from time to time we have also pointed people to it here and said "here, use this"

moooin!

alastairp: yes, afaik there are no direct users of the datasethoster except datasets.lb and labs.api. however, there may be external users of the labs api json endpoints.

i'd only be concerned with preserving the json format unless someone complains. so i think making changes to the UI is fine.

lucifer: yup, right.

agreed.

fwiw, the change i proposed maintains backward compat

oh, so this only changes the frontend?

oh great - I didn't catch that

it changes both but new format is opt in.

that's perfect then, I have no problems with a change like this, let me quickly look at the code again

I was SUUUPER stoked to see my daily jams land at midnight.

there's one incompatible change in the PR which mostly affects the UI. the summary field.

lucifer: is deleting yesterday's daily jams playlist on your list of things to do?

mayhem: from spotify?

also, my daily jams had only 14 tracks in it. Is that because I've been listening and we're scraping the bottom of the CF barrel?

lucifer: yes.

lucifer: right, I'd ask mayhem about the summary field, it might be something that he thought was useful, but maybe no one is using it yet

because if a user walks away for a few days and comes back 2 weeks later, there are 14 playlists piling up.

I think we should make any of the datasethoster changes we feel like.

if someone complains, then we know we have users. then we can be more careful. :)

^

yes agreed. however, my question was are we using that somewhere so that i can go fix it to make it work with the PR? :)

it doesn't sound familiar to me, I don't think i'm using it

let's wait until it breaks for us, then I'll complain to you :)

mayhem: yes makes sense to implement deletion. but to store the spotify url in db first. so next week probably.

alastairp: sounds good 👍

lucifer: were you looking into the metrics stuff on LB to integrate it into the new stuff we discussed at the summit? do you need any help there?

mayhem: re 14 tracks, that probably because you listened to daily jams of the last 2 days. we take top 100 tracks from CF. ~50 of those those get used up first day, the next day another 100 tracks come in but a lot of those will likely overlap with previous day's top 100. the recent listens filter will remove those tracks from the list. so less tracks.

one solutoin as discussed yesterday is to take top 200 tracks instead of top 100. optionally try to use the top 100 first and then if you fall short go for the 100-200 range.

alastairp: i haven't started work on that so far. maybe let's discuss, divide tasks and do it?

ok, sounds like what I was expecting. :)

lucifer: yes, let's do it

alastairp: cool, i'll be back in ~10 mins to discuss.

alastairp: iirc, we wanted to add an endpoint to metrics writer that would expose various metrics to prometheus. i was thinking we could try reusing https://github.com/prometheus/client_python for that.

lucifer: yes, atj pointed out the client to us. the idea would be to use that as much as possible

in the main while loop, call that library to record metrics retrieved from redis. https://github.com/metabrainz/metric-writer/blo...

yep

I recall a few other things discussed there (maybe another level of division? something about data types?) but tbh I was pretty sleepy by that stage of the weekend

so the data type in question here is the metric type iiuc.

counter, gauge, histogram, summary.

i think we can add an arg to BU metrics function to accept the type. by default make everything a gauge.

I think we had a few cases to support both counter and gauge

right, we can port metrics from LB incrementally to the appropriate type.

oh, I see., you mean that the ones we have at the moment make sense to keep as gauge, then we can add counters as needed?

right

ok

zas: atj: what's the story with prometheus at the moment? Is there an instance runnning which we can use, or is that still in progress?

alastairp: there is an instance running on moroder that is hooked into Grafana (stats.)

as we discussed, prometheus is a pull based system, so if you want something to be monitored a new target will need to be added to prometheus

great, so we'll make a start on our pull endpoint and let you know when it's ready

"tbh I was pretty sleepy by that stage of the weekend", are you trying to say that discussing prometheus metric types wasn't stimulating enough?!

is there not an indentation issue here? https://github.com/metabrainz/metric-writer/blo...

lucifer: I'm going to deploy ansible to more of the Spark servers, are you around for the next hour or so?

atj_mb: the cluster is likely busy atm. best to wait for an hour or two

ok

Sure aerozol ! I’ll do so once I reach home. Might take a while

atj_mb: re the code, the indentation is correct probably. because there's a return at the end of try.

but the code probably be cleaned up to make it clearer.

atj_mb: also, just checked cluster is now free. feel free to update the servers.

Hi akshaaatt, I most probably be visiting Delhi on 30th Oct. if that is not possible for everyone then we can meet later someday as you suggested :)

alastairp: did you start working on the metric writer thing?

lucifer: nope!

ah ok, i just realised another thing. we write the metrics to redis in influx format. we'll have to convert those back to python data in metric writer/

ah, true

or update BU to change how the data is written.

if we want to run both in parallel for a while we could decode them again in the writer

then when we remove influx go and make the change in BU