#metabrainz

[musicbrainz-server] mwiencek opened pull request #1953 (master…paged-relationships): MBS-1178: Relationships table paging https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] mwiencek merged pull request #1947 (master…compare-date-period-changes): compareDates / compareDatePeriods changes https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] reosarevok merged pull request #1936 (master…MBS-11408): MBS-11408: Clarify Edit Note Author edit search options https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] reosarevok merged pull request #1933 (master…MBS-11405): MBS-11405: Don't show ended URL rels in DeprecatedRelationship reports https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] reosarevok merged pull request #1931 (master…MBS-10076): MBS-10076 / MBS-11403: Friendlier error when deleting attribute that has children https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] reosarevok merged pull request #1928 (master…MBS-11402): MBS-11402: Block smart links: many.link https://github.com/metabrainz/musicbrainz-serve...

Moin !

mo’’in’

Good afternoon

moooin!

bitmap: for MBS-11409 - do we have a "replace linkedEntities" ticket we could link it to?

MBS-11409: Recurring warning in server logs for recording's page https://tickets.metabrainz.org/browse/MBS-11409

[musicbrainz-server] reosarevok opened pull request #1954 (master…MBS-8232): MBS-8232: Support more Unicode quotation marks in guess case https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] reosarevok merged pull request #1938 (master…MBS-9987): MBS-9987: JSON-LD: Use "Concept URI" for Wikidata https://github.com/metabrainz/musicbrainz-serve...

ruaok: do we want to make creating dataframes and calculating user similarity two different steps or should a request for similarity implicitly create the dataframes?

iliekcomputers: ^

can creating dataframes be a common step for recommendations and user similarity?

in which case, separate, clearly.

yes but i think the time periods will almost always differ

we are thinking in years for similarity but weeks for recommendations

ok, so data reuse is not likely then. but code reuse -- is it the same code at least?

yeah, i am doing that. code will definelty be reused.

the lets keep it as two steps to keep things similar to how recs are being generated.

makes sense

ruaok: https://github.com/metabrainz/listenbrainz-serv... is ready for review.

deep in the middle of that as we speak.

oh awesome.

I'm pondering the URL "/create-user-recommendation/recording" right now.

doesn't seem very RESTy.

/1/user/param/timeline/create ?

/1/user/param/timeline/create/recording ?

yep, happy to change it, i didn't give it much thought tbh

k, let's take a moment to reason it out.

/1/user/param/timeline-event/create

let's first decide the base URL.

`/1/user/<user>/timeline-event` makes sense to me.

yeah, me too.

`/1/user/<user>/timeline-even/create` to create an event. I personally think the type of the event should go in the POSTed data and not into the URL.

+t on that.

thoughts?

the type of the event leads to different types of metadata, which is why i was tending towards different endpoints.

recording -> recording metadata, troi created a playlist -> playlist metadata

ah, yes. good point.

`/1/user/<user>/timeline-event/create/recording` and `/1/user/<user>/timeline-event/create/recommendation` ?

yeah, that sounds good to me.

k

Meanwhile the UI is moving along nicely https://usercontent.irccloud-cdn.com/file/7IjfT...

Mr_Monkey: yes, indeed!

very nice

I put the event time on the right in the end despite my initial reaction yesterday, otherwise the mobile view is really going to be a nightmare

shivam-kapila: you were right about the time ! :p

ruaok: we didn't consider one thing.

i want to keep these endpoints in a separate blueprint.

so it'd be great if the prefix could be `/1/timeline-event`

but its an action done for a specific user, so I feel that it should be under /1/user/<user_name>

hmm.

otherwise this API endpoint is not consistent with our other user focused endpoints.

yeah, thinking about it, you're right.

i'll change it.

I'm reading tests now, will submit review soon.

`Token plsnohack` lol

submitted.

:D

Mr_Monkey: haha. atj also voted for the same. Meanwhile I have a small feedback

The icons on circles look a bit misaligned

In*

(Maybe because they arent symmetric along the vertical)

Yeah, I'm stacking two FA icons (one circle, one icon on top) and scaling them, and haven't gotten to either tuning it or replacing it with CSS

Not sure how consistent the vertical spacing is between icons. We'll see.

Hm. Overall it looks soothing to eyes

Jeez, that like button on my similar artists page is a bit *too* Facebook, isn't it?

I guess it's not trademarked or anything, but maybe we can make it a bit less blatant :D

Why a Spotify Premium account is required to play listens on lb?

there are several reasons. one is a permissions thing that we need that comes with premium, I think. I don't recall the details. and listening to 30 second previews is only a shit experience...

zas: We can only play full songs in a webpage for users with a premium Spotify account. Currently the backup solution (Youtube player) is broken.

Since most of my listens come from Bandcamp, I'd expect to be able to play them from there instead, but I guess this will come at some point.

Bandcamp doesn't offer an API to play music unfortunately

They do provide embedded players though

But no search API. Provided the listen has a bandcamp URI we should be able to play it in the embedded player

Bandcamp URIs are per release 99% of the time

yes, even though that's a bit tricky, as we discussed before, the URI alone doesn't suffice to embed the player, as one needs id

_lucifer: ping

ruaok: pong

hey, I finally have time to do some hacking.

the existing PR implemented most of what I had intended to do, so its not clear what I should work on next.

honestly i prefer listening to music per-album anyway so...

I think currently the issue with Bandcamp URIs in the listens is that we only have the album URI, not that of the song

is there anything on the python/lemmy side I can help with on user similarity?

Ah, that was just mentionned :p

great, I just updated the branch refactor dataframe code for reuse.

Mr_Monkey, zas: if an MBID exists, the MB data might have a Bandcamp link

But similarly, to an album page presumably?

(when MB has a release bandcamp link, a bot could probably fairly trivially link the recordings to the track pages)

not really, i just need to know the format of messages expected by the lb side.

yes, but for the album, not for the track, so we need something to look to the album's containing the recording