#metabrainz

lucifer, mayhem: wonder whether it would be interesting to see which artists somebody listens to in LB that do not yet have genres, and ask them to improve that? :)

I would like to know, at least :D

Also, who should I be sending Android app reports of issues to?

moin!!

<reosarevok> "Also, who should I be sending..." <- Maibi I might . What is it??

Otherwise akshaaatt or lucifer might help

You can direct them to me reosarevok

Ok :) I just didn't know how much time you have to deal with it, but I can do that, will forward

(if you answer, please cc support)

zas: can we delay the Picard release, maybe to Thursday? I'd like to investigate https://tickets.metabrainz.org/browse/PICARD-2515 more closely and maybe have this fixed in this release as well.

PICARD-2515: Swatches not coloured in Options > User Interface > Colours, macOS 12.4

reosarevok: that is not a bad idea at all!

hi, will work from home today

moin and thanks for that.

moin!!

alastairp: I checked the API again yesterday, the user info getting returned. Some data was missing just because the users did not have an mb username.

outsidecontext: ok, np

ansh: right, some users don't have musicbrainz usernames (from reviews that we imported from a separate collection)

For BB, I just need avg ratings to be returned.

which is the other PR?

I raised CB#401 before, but it requires a lot of changes

CB-427: Support entity endpoint in the API including average rating: https://github.com/metabrainz/critiquebrainz/pu...

what changes does it require?

We were adding endpoints for every entity type

So that needs to be done

ah right, because these are entity API endpoints, not review endpoints. got it

But I was thinking if we can send avg ratings at the review endpoint if the user has given a specific entity_id

this is related to CB-437 anyway, right?

CB-437: Add entity metadata to review get endpoints https://tickets.metabrainz.org/browse/CB-437

so within a review block (https://critiquebrainz.readthedocs.io/api/endpo...) we could have an "entity" field, and then inside that the average rating for the entity

will that save an API call on BB?

I was going to suggest the same. Get reviews for an entity, get the average rating as part of the response as well

unfortunately we don't have a great API format to get reviews for an entity and then show entity data

mmm

not exactly related to CB-437. It can be done separately also

oh, perhaps

we _could_ say, only in the case that an entity_id is provided, include another top-level key "entity"

it feels a bit funny if that output format changes depending on the query parameters

Funny but not 🤡

scary like that emoji>

?

Or we can return avg rating here in the review endpoint, and if the user wants the entity metadata also, we can have them in entity endpoints

I think we can duplicate some data in places - so we can have the entity metadata + average rating endpoint per entity

but also add entity metadata to the review (CB-437), and include average rating there too

CB-437: Add entity metadata to review get endpoints https://tickets.metabrainz.org/browse/CB-437

it would mean that if you get all reviews for a single entity then that metadata will be repeated in all reviews, but I don't think that is a problem

Yes, it can be done in this way also

I find the top level key a bit cleaner to avoid duplicating data, but if you definitely want to avoid Schroedinger's response output then duplicated entity info for each review works.

monkey: When will the next BB dump release?

Could also be hidden behind an option

monkey: yeah, I think we considered some inc parameters, but I don't think that this is going to save us much

ansh: dumps are created every monday

okay

alastairp: this is how the response will look after adding entity info in the reviews endpoint: https://www.irccloud.com/pastebin/A1YfYQ7g/

I looks good. I wonder if we should name the entity key the value of entity_type

that is, call it "recording" instead of "entity". I noticed yesterday that this is what the musicbrainz API does when you look up relations

Doesn't that make the API response a bit harder to work with down the line?

I agree. Because the api is dynamic to work for every entity type

given that we already have `"entity_type":"recording",` I don't think it makes it any more difficult

and it certainly makes it a bit easier to define the structure (in pydantic, typescript, etc) of each of the metadata blocks, as we know that they're all separate

for example, see https://musicbrainz.org/ws/2/recording/75c4c85f...

has a list of "relations", and some of them have a sub-item "artist", and others a subitem "work"

makes sense

alastairp: moin :)

Can we discuss the artist conflation issue today? (whenver you get the time ofc)

I no like clowns :c

CatQuest: That statement makes you vulnerable to "Damn you must hate yourself then" type roasts.

That it does :D

P.S. I didn't mean to actually say it. Just wanted to point out someone else might exploit the vulnerability 💀

Pratha-Fish: hi, just finishing up end-of-month paperwork from last week. what a day. I just need to send one more email and then I'll be with you

it's not a vulnerability because i am not a clown. i am a cat

;)

zas: hi, is it possible to use gateways config to redirect an http request to another location (permanent redirect)?. I wish to redirect http[s]://ftp.acousticbrainz.org/* to http://ftp.eu.metabrainz.org/pub/musicbrainz/ac...

I guess this can be done with a custom nginx config stub in upstreams-auto directory in openresty-gateways repo?

yup

but first we need to point ftp.acousticbrainz.org to gateways

have a look at https://github.com/metabrainz/docker-server-con...

it can be done using such config

in this case there's a subpath, I need to check if it works without changes though

it should, but we need to take care of ending /, see https://github.com/metabrainz/openresty-gateway...

wait what's the thing picard uses now. it's not libtunepimp anymore is it? it's called other thing

alastairp: https://github.com/metabrainz/docker-server-con... (but we need a DNS change along this)

CatQuest: uses for?

uh.. idk, it's lib

it has a lib called libsomethingsomething

used to atleast

for..

uh

i dont know :(

CatQuest: you mean mutagen?

I actually wanted to know because on wikipedia there is actually a page called "libformat" for ffmpeg or something, and I wanted to see if something existed for picard's file

hmm

.. no mutagen wikipage is not that :D

libavformat actually, but whatever

libtunepimp doesn't exist on wp

I don't know exactly what you are looking for or what question you want to answer / problem to solve :D There is no libpicard or something. For reading / writing metadata in audio files there is mutagen

only on https://wiki.musicbrainz.org/History:libtunepimp

ah, tnhe I'm just outdated severely. (i remembered the "libtunepimp" name (and it's what I associate "libxxx") with)

yea

lol, libtunepimp.

i mena i remember he name :D

that never took off, lol.

it didn't? huh

but such a cool name

It had been basically become outdated even before I contributed my first code to Picard, and that has been nearly 15 years ago :D

huh. my memory is weird

[listenbrainz-server] 14staticssleever668 opened pull request #2049 (03master…fix/youtube_short_link): Simplify the code to get Youtube URL from a listen https://github.com/metabrainz/listenbrainz-serv...

alastairp: Ahaa, first person asking the real questions about our canonical music_service domain name :D -> https://github.com/web-scrobbler/web-scrobbler/...

zas: thanks for the redirect configuration

mayhem: can you update DNS for ftp.acousticbrainz.org to point to gateways? you can restore TTL to something more sane too

monkey: interesting. this comes back to our discussion about "should it be the domain of the music service, or should you generate it from the url that you have", right?

Yes. The subtlety here is that in JS the browser location will have the full domain (i.e open.spotify.com) but there's no good reliable way to get the root domain (spotify.com) from it (without fetching a giant list of TLDs somewhere)

yes right, and the whole idea of webscrobbler is that it's supposed to be generic and magic?

in which case having a list of TLDs requires keeping it up to date

bandcamp specifically is [bandname].bandcamp.com, right?

In most cases I can think of

That or without the subdomain: `https://bandcamp.com/mr_monkey/feed`

https://github.com/web-scrobbler/web-scrobbler/... this is the list of sites, right?

But I think if you're gonna go and modify each connector they have, might as well hardcode a value

Yep, lots of connectors

it seems that they work by the presence of a particular selector? rather than the domain that it's running on

which is a shame

No, I think they initially work on URL: https://github.com/web-scrobbler/web-scrobbler/...

alastairp: done

monkey: ahhh, missed that

monkey: I mean, my initial gut feel is that `matches` is really close to what we want

mayhem: thanks

and in fact 'id' could be music_service_name, in absence of any better solution

alastairp: Somewhat, but then they also support multiple TLDs sometimes: https://github.com/web-scrobbler/web-scrobbler/...

yeah, I saw a few examples. especially in bandcamp

mmm yeah, that yandex one looks difficult. because in this case "radio.yandex.kz" is probably the correct example, but in the case of a match of *.bandcamp.com we'd want -> bandcamp.com

I feel a bit bad recommending that they add another 500 lines to that file... :)

Yeah. That is something I brought up while we were discussing, but I suppose the argument was "There will be a way to do it, even if it's cumbersome"

Which holds true

Here's a canonical URL where the subdomain is required: reddit.musicplayer.io

musicplayer.io -> 404

yeah, right. I I didn't think about this idea of multiple localised domains that might be different when we came up with the idea

alastairp, monkey: isn't that easily solved on the server side? Just having a rule to drop subdomains from e.g foo.bandcamp.com

and of course the original idea was to try and come up with a uniform way of identifying a site without relying on people typing text which might be different from app to app

IIRC we avoid doing any manipulation of the incoming listens

even if you just do it for analysis reasons that would be possible. or automatically add the canonical version of the input