#metabrainz

CatQuest, it should now be Thursday in your time zone, so let me be one of the first to wish you a Happy Birthday!

rdswift: thanks!

:D

🍰!

I assume you already got the MB cakes, but still :D

goldenshimmer, mayhem: of course, if we *do know* the format, we could still add it :)

(assuming that's what this was about)

moooin!

[listenbrainz-server] 14amCap1712 opened pull request #1776 (03master…validate-listen-): LB-1035: 500 error when sending an array to /1/submit-listens https://github.com/metabrainz/listenbrainz-serv...

monkey: I dont remember what you and i were looking at yesterday, but the release_mbid in playlists is actually not trivial. a few hours of work, but not trivial. which means I can't really dig until the playlists are done.

morning

lol this non-dict breaks the code to jsonify the exception too `TypeError: cannot convert dictionary update sequence element #0 to a sequence`

lucifer: those listens are still hanging around, so my account must be the edgiest of edge cases

nbin: not an edge case but i had discovered another bug in listen counts yesterday evening :(. we are going to rework listen counts so that these bugs go away, listen counts are flaky till then.

no worries. if i add listens in the meantime i assume the count will eventually reflect properly?

yes, when the counts are recalculated the correct one will be reflected.

that's all that really matters then. just figured since i was "starting fresh" that i should probably try to start from 0

maybe i'll use the "-1 when deleting an invalid listen" bug to get there :P

mayhem: Understood. So not a "one line change" then :p

more like a weekend project.

well not quite.

So about a week?

lol

I just need to make a new labs api endpoint for this.

Joyeux anniversaire CatQuest !

and in the process I also figured out how to make canonical recordings, which is something that MB and BBC has desired over the years.

effectively, if you roll in with a recording_mbid, it will look up the canonical one and return it and the associated metadata.

including release.

[listenbrainz-server] 14amCap1712 merged pull request #1774 (03master…LB-1033-fix_doc_comment): LB-1033: Fix doc comment for listenbrainz.webserver.views.api_tools.MAX_LISTEN_SIZE https://github.com/metabrainz/listenbrainz-serv...

[listenbrainz-server] 14amCap1712 merged pull request #1776 (03master…validate-listen-): LB-1035: 500 error when sending an array to /1/submit-listens https://github.com/metabrainz/listenbrainz-serv...

Happy Birthday CatQuest ! 🎉

I managed to make that graph a bit better, thanks for the feedback riksucks and tandy1000 https://usercontent.irccloud-cdn.com/file/xXJYs...

akshaaatt monkey alastairp lucifer tandy1000 reosarevok : if I have made you a playlist (via troi-bot) in the past couple of weeks and you want to save your list, clone it soon. I'm going to nuke all generated playlists soon.

!m monkey

You're doing good work, monkey!

Thanks for the heads up

Definitely want to save one of those playlists

Already done ruaok 👌

yeah, I'm saving my missed tracks one.

mayhem!

:D

mayhem: 👍

python logging is so confusing

alastairp: agreed. one of my least favorite parts of python

and it's so opaque in so many places, you have no idea _what_ is doing the logging

I think I've finally got the double-logging stuff fixed in both dev and prod

unfortunately, I now think my recommendation from earlier in the year to use `current_app.logger` everywhere was a stupid move and doesn't make sense, so got to put that on the future todo list again

why so?

flask has a single logger, named the thing that you pass in to the `Flask()` constructor. in our case this is `listenbrainz.webserver`. We print the name of the logger in log messages:

> spotify_reader_1 | 2021-12-09 12:12:19,994 listenbrainz.webserver INFO Processed 1 users successfully!

this means that a log message coming from `listebrainz/spotify_updated/spotify_read_listens.py` still has a name of `listenbrainz.webserver`

not the end of the world, but makes it a bit more difficult to work out where the message comes from

clear as mud. :)

I was just confused why _all_ of our log messages appeared to be coming from the webserver package

outsidecontext: there is something weird going on here https://i.ibb.co/BwjkBZr/Screenshot-2021-12-09-...

Picard appears to be changing electronica to electronic. I just checked the tags and it's written electronic twice.

atj: how have you configured genres? is this genres from MB (the built-in ones in Picard options), or any plugins?

lucifer: I think I've finally fixed https://github.com/metabrainz/listenbrainz-serv... now, and https://github.com/metabrainz/listenbrainz-serv...

let me futz with genres a bit more, and then if I have time I'll get back to https://github.com/metabrainz/listenbrainz-serv...

nice

alastairp: btw, do we now have the CB review filter to get only text reviews?

When are you coming to the motherland reo? Planning on stopping by here?

Nah, Madrid / Asturias only this time

Makes sense. You'll be missed !

Next Autumn, maybe? :)

That's for seeing you, not for the alastairp code

(but maybe both :D )

outsidecontext: those are genres from MB. Configuration: https://i.ibb.co/Q6X6rYX/Screenshot-2021-12-09-...

and what release is that?

reosarevok: CB managed to find itself on the backburner again :( Maybe lucifer and I can do a hack week after YIM is finished or early in the new year to finish up all of the PRs (and the new LB mini review functionality!)

no plugins installed except for a custom one I wrote to define some script functions for my folder name conventions

sigh

alastairp: ok, no worries :)

the release is https://musicbrainz.org/release/72b23cdf-493c-4...

(just wanted to make sure re: the MB PR)

that's odd. does this only happen if you have files on that release, or does it show up the same when just loaded?

I just loaded the release manually by entering the MB URL in the search box and it shows electronic twice

I feel like this is too odd, so it must be something I've done :)

can't reproduce this here. a script maybe?

for me it shows only a single genre, "Electronic", which would be expected with the given settings (because of the minimal genre usage threshold)

alastairp: when do you think you'll leave the office today? because the first conversation is really more a request than a long discussion...

outsidecontext: I'll investigate further and let you know if I find a bug

alastairp: looks like I might return closer to 3pmish.

alastairp: 1765 lgtm but something in frontend tests needs poking. https://github.com/metabrainz/listenbrainz-serv...

mayhem: I think we'll do lunch and then I'll be around until 4ish, so should be fine to talk then

lucifer: good catch, thanks. urg, this will be due to /code being owned by root. will have to rethink it

maybe just preventing pyc files is a good enough fix for now and we can look at it later

sure sounds good.

alastairp: 1773 looks good but will probably need a change. i realised after approving that load_config should probably come before logger init.

+1

urgh. I thought that I put it there, I must have been moving it around while testing stuff

ah, that's right. I removed DEBUG from my config file and was using the FLASK_ENVIRONMENT env var to set it

so yeah - if for some reason we set FLASK_ENV=production, but config.DEBUG=True it wouldn't necessarily work as we expected it to. will change

<monkey> "I managed to make that graph a..." <- lookin good

<monkey> "I managed to make that graph a..." <- graph is looking good! what helped?

tandy1000: Removing values on the bars, and figuring out how to massage the data to get a better looking X axis

alastairp: ready to chat when you are.

just debugging something

mayhem, you might want to update your IRC nick on https://metabrainz.org/team

I did the code change, but haven't deployed yet.

👍

mayhem: https://twitter.com/colinraffel/status/14686188...

yvanzo, bitmap: updated https://github.com/metabrainz/musicbrainz-serve... with the suggestion to allow reuse, would be good to get this ran this year :) Maybe next week even?

lucifer: I just tried to deploy metabrainz.org and I am getting:

anything you've encountered before?

alastairp: https://github.com/bookbrainz/bookbrainz-site/b...

mayhem: haven't seen myself but looks like https://stackoverflow.com/questions/70068407/mo...

I'll go deal with this in the new year. too much YIM left to be YAMmering about with this.

quick look at https://github.com/metabrainz/metabrainz.org/bl... don't see us pinning the version so unless a dep show does. possible it pulled in a latest version which changed some imports

if you want, i can look into it later today or tomorrow?

sure, if you feel inclined.

👍

also, let me know if i can help with the left yim stuff

k, not sure if there is a chunk that can be broken out. I only have submitting playlist excerpts to LB left.

invader yim: https://static.wikia.nocookie.net/characters/im...

what's the plan? troi dumps a csv, we import the csv in LB with a script?

I dislike the repeatability of this plan, so I am going to add a troi element that will query LB directly to fetch a list of users who do not have YIM playlists for a given troi patch. and the end of the pipeline will be a submitter element which executes an UPDATE statement on LB to insert the JSPF except into the DB. should the process get interrupted, I just start it over again.

monkey: ` "test:ci": "JEST_JUNIT_OUTPUT_DIR=/tmp TZ='Europe/London' jest --ci --reporters=default --reporters=jest-junit",`

that begs the question -- in which order do we process these things?

if I do an UPDATE, then I am assuming that most of the data already exists and that I just need to add one key value pair.

order of processing as in order of other YIM stats and playlists?

yes

to confirm, troi will create a playlist for the user and then add a link of it to YIM table with a specifc key?

I expect that generating playlists might take as long as a day.

lucifer: yes, to be specfic, monkey requested that I upload a 5 track JSPF except file to LB.

excerpt

i see. i would say we execute both together, and let INSERT ON CONFLICT DO UPDATE do its job. all the stats jobs are written that way.