#metabrainz

right

ok, then it makes sense to me.

lets do that.

👍

We've taken the hard decision to shut down the AcousticBrainz project. Read more about why here [url] - good for twitter?

project, because of low data quality. Read more ....

?

yeah, could do

I investigated the OCSP issue, it is due to a bug in autossl, related to caching, it is cached for one hour, so the issue will only happen once in 3 months (on cert renewal), only on FF, for max 1 hour. That said, I asked autossl devs about it, if no patch appears in near future, I'll try to provide one, though I'm not really at ease with this stuff (lua + ocsp).

would a fix for this issue carry forward to the new gateways, zas?

monkey: https://github.com/metabrainz/listenbrainz-serv...

your comment `You'll get a 404. Changing the URL from /recording/ to /track/ gets a hit.`

oh, maybe never mind.

Only on album pages, not for recording_mbids=…

release pages*

ok, i see what you're saying now. I wasn't certain which URL you were referring to.

mayhem, monkey: took a brief look at the PR. i haven't tried it yet but does spotify playback work for you when using it?

now that you mention it, no.

lucifer: do you still have the gists you posted about the BU data being returned?

Hm.

No

ok that confirms my suspicion, you'll need CORS headers on the endpoints otherwise the browser won't send cookies and you won't be logged in and no spotify token.

in one case it appears to be using track_mbid, not recording_mbid. Maybe both fields are present?

ok, let me add them.

but using the stock CORS headers we use in API won't work either because '*' in allowed hosts prevents cookies too. and there are another couple of things to confirm.

I don't think it's only the new player endpoint though. On test.LB on my feed Spotify is also broken:

`@crossdomain(headers="Authorization, Content-Type")` ?

"Spotify error: Cannot refresh Spotify token right now"

monkey, but does the player endpoint even log you in?

mayhem: well, yes, since we'll likely not use autossl anymore

Ah, well I signed out to try, indeed no sign in

lucifer: I didn't require logins, because I wanted it to be as widely used.

but I suppose we can add them if need be.

mayhem: no, that likely won't work. i can look into it later today or early tomorrow.

lucifer: ok, please do.

mayhem: yes that is fine. i am not saying require login but if you are an existing LB user and have spotify connected, it'd be nice to use spotify.

zas: then I would say don't fix it yourself. just be ready for the next time the cert gets renewed, rather than spend time for a fix.

thx!

re BU gists, I have a few. this is the latest one i had posted IIRC.

I want recording_id, not mbid then. easy fix.

lucifer: I logged in again, Spotify does work on the new player endpoint for me

(I think I was testing with an album not in spotify)

I am logged in and the player says: "You need to connect to a music service and..." which I have...

I see the same message, but when i then play a track it does work with spotify

ok, promising.

but why the message?

A very good question indeed.

monkey: the track mbid issue is now committed.

weee !

lets update test at some point to confirm that, please.

Can do

that's... interesting

a significantly larger number of unique tags applied to releases compared to recordings

that being said, they're all count 1 and all garbage values

(mb tags)

not surprising to me. I'd more inclined to tag releases than recordings

not very useful tags, though

mayhem: yes that should be simple to add. not at my machine currently but can add that later.

monkey: so working but wrongly shows error message?

monkey: https://musicbrainz.org/release/9a193023-94ae-4...

mayhem: sorry., back now - just reading the email

I always recall the worst bit of data in lastfm was that some huge unknown number of "recording mbids" were actually track ids

lucifer: Definitely something not working right, but I'm not sure what. Player does work, but initially does not take user's spotify auth details into account

I can't remember what mlhd data looks like. do you have it somewhere>

mayhem: I updated test.LB, recording links now work fine

sweet!

ok, that PR is done from my side then. I'll leave it in your hands monkey.

Plenty of hands on a monkey.

you dirty monkey.

I think we should merge then, and keep it low-key until we have better metadata to show.

But it's a good starting point

ha someone approved it?

has

Negative, I spoke too quick. Once approved it should be ready to merge

I'm happy to keep it on the down low for now. as long as I can use it.

lucifer: monkey : https://bono.metabrainz.org/recording-similarit...

"open results as playlist" \ö/

[data-set-hoster] 14mayhem opened pull request #4 (03master…add-playlist-support): Add playlist support https://github.com/metabrainz/data-set-hoster/p...

Sweet mayhem !

akshaaatt: I'll work on adding some fontawesome types. can you squash the commits together in the meantime?

Sure bitmap, sounds great! Thanks

Sure bitmap, sounds great! Thanks

zas: outsidecontext : Can you please take a look at the Picard ideas page for GSoC and make sure it is up to date?

mayhem: yes, will do so tomorrow

outsidecontext: thanks!

alastairp: would adding BPM support to Picard be a possible GSoC project?

yes, I think so, though we should discuss it with the picard team to work out the best way to distribute it (e.g. one of the recent tools requires all of tensorflow, not sure the best way to distribute that)

and we didn't come to an agreement yet on if we wanted a more advanced music player

ok, lets have a chat about this project in general. maybe there are 3 algs that don't need tensorflow...

to clarify - I make it sound like tensorflow is a big deal, it's not really - it's pip installable. I just don't know how that ties in with the picard app/plugin distribution model

ah, ok.

lucifer: how about the content resolver, at at least the metadata cache part of it being a GSoC project?

alastairp: in general we can add that dependency. depends a bit how much this adds to the overall size and if it is compatible with all platforms. also ideally it becomes optional so it is easier to handle for users who don't want to install it (e.g. on Linux distributions)

outsidecontext: great. it looks like it's available for win/mac/linux just fine: https://pypi.org/project/tensorflow/1.15.5/#files, but is up to 130mb in size. optional sounds good

also there is https://github.com/metabrainz/picard-plugins/tr... . no idea how well this performs. it uses aubio

what we don't have easily is a way to bundle dependencies with plugins

130 MB is huge, but maybe this can be added partially?

So it looks like I lost my password to log in to the MB wiki, and the lost-your-password page doesn't seem to send emails. I've tried with name and/or email and waited a couple of hours, but did not receive any email.

Who should I ask about this?

outsidecontext: actually, looks like we might be able to modify it to use tflite

monkey: sounds like there are a handful of wiki things that need to be addressed, so we should try and pin down our wiki wrangler

https://pypi.org/project/tflite-runtime/#files 2mb

+1

Can we move the ideas page to a GitHub readme on some repo called metabrainz -ideas?

We have a lot of GSoC-related info in the wiki that we link to, it would be cumbersome to move everything. Is it just for ease of editing?

I think the GitHub UI would ease our work managing this part of the wiki since over here all we need to do is present the relevant ideas and proposed mentors.

Is there anything in the wiki which can't be moved to a readme?

I'm not proposing we do it, but it can be a good decision if it isn't too hefty of a task

Well, for starters we have the pages from previous years' GSoCs: https://wiki.musicbrainz.org/Development/Summer...

And related ideas pages for each year (or some years)

Not sure Github would really bring us much that we don't have on the wiki, while creating a good chunk of work

Agreed. But the wiki UI strains my eyes since we have a lot of reading to do and the font is too small.

How can we edit the wiki codebase? Is there room for making this change? I could look into doing this

Either we look to update the wiki or remove this hassle for us slowly by moving yearly work to GitHub/any other free platform in the coming years and link the older stuff to the wiki for previous years ideas, etc

hi mayhem alastairp ,

I have some feature suggestions for LB, should I post it here, or in the community, or here first and then the community?

riksucks: a quick summary here would help us decide. :) but opening tickets is never a bad idea

mayhem: so what I had in mind was, in LB, when you hover your mouse over a listen, you can see a play button appear. Similarly, why not add a (i) button, when clicked, would show... (full message at https://libera.ems.host/_matrix/media/r0/downlo...)

I believe, matrix mangled my message due to this being multi-line, so posting it again

1) name of the song

2) artist

3) link to go to the song in MB (if exists)

4) how many users have recommended, pinned and hearted this record

5) 2 tabs, CB reviews (default tab) and Pins (inactive tab, when clicked, would show all the texts of the users who have pinned the record)

basically the idea is to give every record it's own page, some details about it, and what people think about it (via CB and other LB metrics)

rik: could you mock up how this would look?

i cant imagine it

sure, I would send it by morning (5 to 6 hours later) :D