#metabrainz

no sabía que había tantos españoles por aqui!

por supuesto, sí!

El español de todos estos no va muy allá aún, pero están trabajando en ello, como Aznar y Bush

No, el único español soy yo, que fue el que salió de España corriendo :D

ah jeje

mayhem es alemán, Freso danés, donalastairp de Nueva Zelanda (si no recuerdo mal), Gentlecat es ruso, y Quesito americana... creo :)

sí, reosarevok.

Sí.

Si

:)

> Last.fm currently shows top artists, top albums and top tracks on the user page, with the flat list of listens that ListenBrainz already has. On artist and album pages, it shows top listeners and top tracks prominently.

the first sentence is confusing. it mentions both last.fm and listenbrainz.

I think one thing we should do is explain in much more detail what each of these graphs is supposed to show.

there are 3 classes of graphs we're going to have: per user stats -- what did the user listen to, which artists/releases-groups were popular with that user.

and then entity pages (artists, release-groups) that show how popular that artist is overall and which users are top listeners.

and the third being other things that are neither user nor artist centric. I can't think of any examples right now.

I'd like to see these graphs described in more detail, please.

mayhem: maybe an overall genre graph of the site entirely i.e tag popularity on the site, that could fit in the third category

sure.

so, that is your first todo.

second, lets chat about bigquery.

Okay, I will try to get some images and more descriptions of the graph

mayhem: yes please

there is a lot of uncertainty around this just now. we simply do not know how to best do this yet.

but, I think a first cut should look roughly like this:

create a set of scripts that get fired off from cron periodically.

these scripts wake up and calculate the stats for a set of users.

the results should then be stored somewhere -- could be PG, could be redis. we need to discuss.

where somewhere really means "on our server"

then we serve/render the results from our server.

Redis would be faster, I think

redis has a direct cost implication, possibly.

BiqQuery is free for the first 5TB of traffic. We will get more free traffic from Google, but let's assume that our user of BigQuery won't be fully free. there may be some cost associated with it.

so, if we spent $20 to calculate monthly charts for a pile of users and then we need to restart redis, then what happens?

we have no graphs to show to our users in the meantime and we're out $20.

with me so far?

yeah, that is bad :/

not terrible, but it speaks more for postgres, I think.

mayhem: redis can be made persistent though, right?

we need to make redis persistent as well, so that might be an option.

:)

this topic would be good for you to think about more.

but, you can see how calculating stats becomes a batch operation, no a live operation.

I envision calculating monthly stats for anyone who logged in in the past month.

we'll just have them ready, calculate them on the 1st of every month. or maybe every monday.

just a matter of when we run the cron jobs.

now, what if an established user comes in after not having visited in a month?

every week at least would be more preferable to me, but it depends on the money we have I guess

money, time, google's generosity, lots of things.

mayhem: we put her in some kind of queue, i guess

really hard to predict right now. let's make it flexible so we can adjust as needed.

yep, a queue.

show the old graphs with a fat note: hi! Welcome back. We haven't seen you in a while, so we didn't compute your graphs. we're doing that now. they'll be ready shortly.

once they are ready and the user reloads, we show updated graphs.

sounds good to me

does that give you with which to plan your proposal?

mayhem: one more minor implementation detail about the graphs

shoot

Is it okay if we use some charting library based off d3 or would you prefer to roll our own charts using vanilla d3?

I came across plotly after some searches and it looked okay to me

I use plotly myself for personal use.

I like it, but it is a paid solution, no?

I would prefer to stick with just a set of libraries, wether plain vanilla, derivative, or plugin enhanced.

plotly.js is an open source graphing library

ah, I wonder if there is any connection to: https://plot.ly/

yes, there is. ok.

same guys made it

really, any open source library should do fine. but this is a decision I would like to get community feedback on.

see if people have used the various libraries and if there are things for us to watch out for.

okay, I think I'll make a post on discourse about it. people on irc can chime in here also

good plan.

we can run scripts for artists and albums etc also, but I'll have to think about how to choose which artists we generate when.

yep.

[musicbrainz-server] reosarevok opened pull request #460: MBS-9176: make Create RG form consistent with others (and less ugly) (master...mbs-9176) https://git.io/vDwbr

mayhem: thanks for the help, I'll get back to you with a more complete proposal in a few days hopefully :)

fabu. I'm soo looking forward to this. :)

Yay proposals

\o/

!m iliekcomputers

You're doing good work, iliekcomputers!

Freso: https://tickets.metabrainz.org/browse/MBS-8820 is listed as GCI, so do you remember if we had decided what of the two options to implement while adding the task too?

MBS-8820: An artist with only event rels has a fully blank relationships page

I'm not sure the task was added at all.

I see the "GCI" label as a way to find GCI task suitable tickets, so it might just have been deemed GCI suitable, but since it isn't currently actionable, not added as a task?

[musicbrainz-server] reosarevok opened pull request #461: MBS-9166: make guess case ignore "Early" unless already in parens (master...mbs-9166) https://git.io/vDwAM

Whoa. Look at reo go!

There'll be no more open MBS tickets left soon!

:p

reosarevok: No option had been decided.

Ok :)

GCI labels on MBS tickets were for actual GCI tasks, not just suitable ones.

yvanz: That's how you used them, sure. Gentlecat and I used them for GCI suitable tasks, so that they're also easy to find in 8 months.

But you did not use it for MBS project.

Labels are not project-specific.

Freso: That was just an attempt to provide an useful answer to “it might just have been deemed GCI suitable, but since it isn't currently actionable, not added as a task?” I did not comment the way you use it.

reosarevok: no sabes como decir neozelandés?

donalastairp: nunca recuerdo si eres de NZ pero viviste en Canadá, o al revés :)

that's the one

I was just pointing out that for me you said the country name, while everyone else you said their nationalities

Oh

For some reason it came out like that. I do know the nationality, but it's not the first thing in my head, maybe because it sounds a bit weird :)

yeah

for me "soy de nueva zealanda" is much easier to remember and say

Freso: how do I split https://community.metabrainz.org/t/musicbrainz-... plus my reply into a new topic?

reosarevok: On the right side at the top of the view (ie., it moves along with the scrolling), there should be a wrench icon. Click that, "Select Posts...", select the posts, then "Move to a different (new) topic" (or something like that).

Thanks!

No problem reosarevok! No preosarevok!

reosarevok: ... Soundcload? https://community.metabrainz.org/t/importing-so...

*cloud

Isn't it talking about Spotify?

"I've been thinking about soundcloud as they have an api."

Oh.

Most sites not living in the 90's have an API these days.

i especially like the oreosaverok. with milk :D

mayhem: Freso: /anyone else with opinion: so for officebrainz logos, are we talking about a whole set? (various detailed lobes, lobes +words)

And any ideas so far? (I have some..)

it would be nice to have a full set, but not sure if that is really necessary.

no real ideas yet. what do you have in mind?

What is officebrainz again?