#metabrainz

Jeez. Does LB really need Google stuff? Because those emails where they're never happy sure do look like a pain in the ass

yvanzo: I've updated PR 2151

that is exactly the question I was just asking myself.

I'm going to mail my contact at YouTube and ask him for advice.

now we are just going around in circles...

i do not understand which point of the list we do not meet.

yeah. all for something we don't actually want. which is the worst part.

exactly.

their last email about Privacy Policy did not have this section.

I think each reviewer we get, applies their own standards.

seems intractable.

yeah :/

oh well. did you see the security assessment BS down that page?

can we request NO SCOPES?

that was how things workedd till last year. google deprecated an removed that.

are there any non-restricted scopes at all?

not for youtube i think but i'll recheck.

morning

ruaok: sorry, I didn't reply to that tweet. just done it now

yesterday wasn't going my way, I turned off all of the computers at midday and took a day off, it was definitely needed

lucifer: agreed, we should structure the training process such that if an error happens, we propagate an exception to a single top-level error handler, report it to sentry, and then log it in a way to show the user that there was a problem

hi!

yes, makes sense.

i made some more progress on the sklearn stuff, i think its mostly ready except for the exception handling part.

alastairp: thanks!

[listenbrainz-server] 14amCap1712 opened pull request #1542 (03master…youtube-again): Query Youtube API using api key only https://github.com/metabrainz/listenbrainz-serv...

thanks param !

[listenbrainz-server] 14mayhem merged pull request #1538 (03master…Pinned-Recording-Modal): Add Pinned Recording Modal https://github.com/metabrainz/listenbrainz-serv...

sorry for the late review, I only just saw it today. :(

all good.

lucifer: another thought I had... I think we need to build content resolution into lb-server.

having it in brainzplayer makes BP far more complicated than it needs to be.

Jesus

We have a week and a half of 30+ degrees coming

Maybe I should get a plane to somewhere cooler, like... the Mediterranean :p

yeah makes sense. so send 1 query to LB server and then LB sends back the link to play?

lucifer: or even on a whole playlist at a time. so that BP can just do playback and nothing else.

reosarevok: Is 24C here. :)

yup agreed.

ruaok: sadly we are only 7 days past the second vaccine, so a bit early to travel anyway :(

and you don't want to come here. a huge number of youngsters, who never cared about covid are getting infected in record numbers.

Well, hopefully the vaccine would help with that... but yeah

[listenbrainz-server] 14amCap1712 merged pull request #1542 (03master…youtube-again): Query Youtube API using api key only https://github.com/metabrainz/listenbrainz-serv...

ruaok: let's release?

go!

meanwhile: Francia aconseja no veranear en España y Portugal

yerp.

[listenbrainz-server] release 03v-2021-07-08.0 has been published by 14github-actions[bot]: https://github.com/metabrainz/listenbrainz-serv...

veranear, what a verb

30 degress? Its 50 degrees here

I can't wait to see what the foreign ministry of germany says for next week.

I'll be coming vaccinated and with a PCR test, but still.

shivam-kapila: I'd like to think your buildings are built taking the heat into account though, at least?

shivam-kapila: reosarevok is a wimp

(if not, yeesh)

Ours aren't, since we're a cold country - they're mostly built to keep the heat in :p

Not at all reosarevok

nice of you to think buildings are planned before building in india

haha

Sorry I had hopes for your country, should have known better (?) :p

your insulated house is good for keeping heat out... for a while.

On point lucifer. Plus mine is a rented place. Why would the landlord care

At least we're "only" at almost constant 27 degrees inside

Some people are at 35, so I guess I should be happy :p

My mac is so heated that I can cook chapattis on it

I'm just glad that evenings still cool to 21C.

It's a more normal temperature in Lapland, so that'd be an option - sadly it's death by mosquitoes in there in the summer AFAICT :D

Do they make laptops you can use while sitting in a river yet?

Then offices will be in swimming pools rather than buildings :p

Sounds like a plan

lol someone at work got github copilot to solve an interview problem we give

Its trending right now

How copilot can help crush leetcode questions

LB prod updated.

<param "lol someone at work got github c"> Oh my😂😂

alastairp: the best model pickle file should be the history path in db, right?

lucifer: right, they're equivalent, but perhaps we could use a different key for the sklearn one, maybe just "model" ?

we could do that but that only causes to add more ifs elsewhere.

whenever we want to access the model. i don't think we do it anywhere currently though.

i'll do the model key then, if we need to change we can do that later anyways.

alastairp: there's also this warning in sklearn logs

lucifer: how many items are in your dataset?

345?

345

yeah, looks like the warning is pretty harmless, then

👍

[listenbrainz-server] 14amCap1712 opened pull request #1543 (03master…revert-privacy-policy): Remove LB specific privacy policy https://github.com/metabrainz/listenbrainz-serv...

[musicbrainz-server] 14yvanzo closed pull request #2081 (03master…MBS-10211): MBS-10211: Formatting Tweet links for recordings https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-android] 14akshaaatt opened pull request #82 (03master…asset_attributions): Asset Attributions https://github.com/metabrainz/musicbrainz-andro...

Hi lucifer !

hi!

<BrainzGit "[musicbrainz-android] akshaaatt "> If you could confirm quickly whether this format is good?

For attrubtions, I have basically prepared a table

attributions^

sure looks good to me. I would suggest to use markdown file instead though so that formatting is easier.

Oh Okay!

and link the master version of this file in the about page of the app.

Cool Sounds great!

yyoung: are you able to run selenium tests atm?

yvanzo: I haven't tried to run it locally since then, I just found that CI combining with local Selenium IDE plugin works fine

Ok thanks, I will look into running selenium tests locally, because it would be nice to add tests for your last fix in 2151.

If I was able to import json5 files into it, then it would work perfectly

BTW, I just found that the saved file of selenium IDE is in json format, maybe I could write a convertion script

yyoung: If you’d like to explain how to use that plugin with CI, we could probably add it to HACKING.md notes.

Currently I have to manually enter every action from json5 to the IDE

If I found a way to convert, I would be glad to write the notes :)

alastairp: i'll start work on the model testing part meanwhile in a differrent PR. this only is already too big. sounds good?

lucifer: yeah, I think that's fine. this is the intermediate tables for doing evaluation?

alastairp: yes, that and in general for external validation of models.

cool

up to you - but we could even do this without the sklearn code integrated for now

in order to prevent us getting into trouble with multiple large prs that merge into each other

yes, right. i'll base the PR from master.

or should i do it from Py3 branch?

cool. gaia only, but designed so that adding sklearn in later would be fine

mmm

the highlevel extractor runs with subprocess.run, so we could probably get away with py3

yeah indeed but it uses future module.

i saw a ticket on gaia repo that py2 futures backport is different from py3 inbuilt futures module.

note that this is during the training process

ideally we'd still keep this in py 2

ah ok.

py2 + six then?

yep

+1