#metabrainz

[musicbrainz-server] 14derat opened pull request #3011 (03master…link_dir): MBS-2604: Infer link direction between people and groups https://github.com/metabrainz/musicbrainz-serve...

derat: thank you for working on that ticket!! Will be extremely useful! (I haven't looked at the code though)

aerozol: same thoughts

But where to use orange is the question

also did you brainstorm ideas for showing similar listens

My thought is to use the same UI component used to show content of cb reviews for this as well

jasje: you don't *have* to use orange at all

so about the similar listens design

We already have it on some icons and visuals in the app proper

With similar listens can you remind me? I believe you had enough to start putting it together, and then afterwards we were going to look at how to show the similarity score

yupp

currently i have it like this

aerozol^

Its missing the time tag and a quick follow button

mostly the time tag

If i add it below the listen, things just look too spaced out and gets out of sync with the padding i have on every list on the app

aerozol: something like this would be good i think. Not sure what to fill the bottom side with. But all in all I already have a component ready for this https://usercontent.irccloud-cdn.com/file/76EXW...

mayhem: do we have any current last.fm contacts to talk to about last.fm users coming to make a mess in MB?

I'm not sure they'll care and/or be able to do anything, but

not really no.

jasje: Looking at this I’m still leaning towards not having the similarity or the follow button duplicated everywhere

mayhem: the listening now tags PR is already mergred.

is it not enabled? or do I just not know how to use it?

it should be working normally.

are you able to tag on now listening? I can't see how to do it.

click on add tag to enter a new tag

hmm actually i get 401 on that.

I dont see the add tag:

is this prod?

omfg. I had to restart the metadata viewer.

aerozol: im not in favour of follow button as well but similarity has to be init

now its there. HOW in the hell?

thats like the whole purpose

but, yeah, tagging errors out.

This seems much more efficient: https://usercontent.irccloud-cdn.com/file/Zagbv...

yup on it

If you don’t like the look we can change it

I don’t like this one

I’m going to ask monkey for a second opinion, but we’ve gone back and forth for long enough that I’m probably going to make a call on it

If you want to give reasons as to what/why/how you don’t like it that will help!

this bar has totally independent state than that of listens list

its like first finding a listen and then matching it with a user in this component

I see that as a positive - as you’ve noticed, myself and a few others see this as a playlist first, and a follow similar users display second. I’m not saying the second part isn’t important, but it’s not necessarily what everyone’s after

What would you prefer - we discuss it with monkey and make a call, or I put the different options on the forums for some user feedback and thoughts (and then me and monkey make a call)

something minimal which can be added as extension to the listen card

i really just want the extra similarity and thats it

rest of the card looks cool

There are a few problems solved by a separate section, with the ten users ordered by similarity: It explains what this ‘similar listens’ tab does/shows, at a glance. And it lets us see the % and click through to follow without duplicating everything.

mayhem: can you logout and login again?

it should start working for a while, and will probably break down when the token expires. will look at it again, have a hunch as to why its happening.

My call right now is that I don’t want the % and the follow button duplicated on every listen, I don’t think it’s good UI. I’ll let monkey chime in as well. As said, happy to get community feedback if you think users will prefer your version

I'm viewing in the app itself -- there are no means to navigate, so how can I log out?

mayhem: login and logout from browser and then restart metadata viewer.

ah.

we should improve metadata viewer UX though in future.

aerozol: the follow button is stupid i agree

!m lucifer

You're doing good work, lucifer!

tags working well!

all monkey's work :)

you worked out how to fix it. but yeah...

the token refresh is still broken but i have added logging on test for helping in debugging, should have more data after 1 hour.

!m monkey

You're doing good work, monkey!

I'm sooo happy right now. :)

> My call right now is that I don’t want the % and the follow button duplicated on every listen, I don’t think it’s good UI

aerozol, jasje : ^this also sums up my opinion. Extending the listencard and taking more space just to repeat information that could be collated at the top just seems like a waste to me (also because I want this to be a way to play and discover tracks first and foremost and the other data gets in the way. I'd rather see it tucked away in it own section

But honestly at this point it feels like we're repeating ourselves. The conversation feels like deja vu.

mayhem: Thanks, i'll note down that greying and disabling the add tag button would be better than hiding it which is probably what happens currently

As for token refresh issues, lucifer will be my savior

monkey: It is a repeat, but last time I was a bit vague on the outcome. This time if you feel the same then I would say decision made

are you going to make another pass at the listening now page?

if I wasn't about to get on vacation I would make a couple of improvements on it. (artist summary and artist info)

mayhem: when will you be back from vacation?

aerozol: Seems pretty clear to me. information repeat is undesirable so I would select another way to show that information. If the one with the percentage gauge with the username underneath (posted above) isn't satisfying to you jasje I think we're open for discussion, but UX-wise I think the two professionals are in agreement

Officially Sept 4. but I'm mostly sort of around this week.

mayhem: Yeh, improvements required

tday - wed, I won;t be around for afternoons.

oh cool, wanted to discuss further tasks other than MeB ouath.

In particular I'm starting to hate that you have to hide some of the info to see the rest. Accordions be gone !

lucifer: I'll be around an hour each day as per my "vacation" policy.

but, we can discuss oauth tomorrow morning. I might also work for a little while this afternoon.

hehe, makes sense.

it is actually a weird feeling. I am so engaged in work, I'd be happy to skip vacation. but I haven't had a proper break since march, so I should really go. :)

yup indeed

mayhem: since GSoC without extension deadline is on 28th, we should ask all students on that timeline to start working on their blog posts for submission.

indeed. a topic for the meeting?

we can then schedule those for publishing between 21-28, the submission week.

(if people bother showing up this time, lol)

yup makes sense

lol

Hey reosarevok bitmap would you guys be free for a discussion regarding the project extension after the meeting today?

Or before. Whatever suits you

I should be around a bit before and a bit after so both work for me

Let's see about bitmap

epic :D

mayhem, monkey: to get the OAuth stuff to work we will need to do 2 things, 1) revoke access of the LB prod from MB website https://usercontent.irccloud-cdn.com/file/sDWlm...

2) log out the user from LB.

i can work on SQL scripts to do that for all users automatically.

lucifer: SGTM. What was the issue with the LB oauth app, out of curiosity?

aerozol: alright I’ll brainstorm stuff based on this idea now

monkey: i am not sure but i think it has to do that we are now requesting extra permissions from the users which were previously not requested.

ahaa

[listenbrainz-android] 14akshaaatt merged pull request #218 (03dev…settings): Settings Page and Cleanup https://github.com/metabrainz/listenbrainz-andr...

lucifer (maybe mayhem ?): thoughts about dropping the snapshot tests in LB frontend tests when we move away from enzyme to use react testing library?

I haven't really been able to find much use to them myself as the diffs are often unreadable. By that point, they're mostly just in the way and a chore I could do without.

[listenbrainz-android] 14akshaaatt opened pull request #219 (03main…dev): Dev to Prod https://github.com/metabrainz/listenbrainz-andr...

[listenbrainz-android] 14akshaaatt merged pull request #219 (03main…dev): Dev to Prod https://github.com/metabrainz/listenbrainz-andr...

[listenbrainz-android] 14dependabot[bot] closed pull request #212 (03main…dependabot/gradle/navigationVersion-2.7.0): Bump navigationVersion from 2.6.0 to 2.7.0 https://github.com/metabrainz/listenbrainz-andr...

Hey monkey, I'm currently working on my GSoC project and the submission of my research project in parallel, which proved to be not that effective.

So I wanted to ask whether it's possible for me to pause GSoC for two weeks, get my written draft and presentation for the research project done, and then fully focus on GSoC again.

Of course I will compensate the missed weeks by appending them to the end of my schedule.

monkey: not sure what the negative impact of that could be. but I'd be happy to see them gone. every time I want to make a simple text edit, it takes me 10 seconds to make the edit and then an hour of fighting with snapshots.

Yeah, that's pretty much everyone's experience as far as I can tell.

Picard uses to update some of its data (the list of media formats and release and release group types) from the attributes.pot file of the musicbrainz-server. I noticed that this has not been working. There have been a couple of additions to medium formats and release types that are not present in this file, which was updated last time in May. Is this intentional?

monkey: has it cauight bugs that would've caused us issues in prod? If not, then there is not question about yanking it.

outsidecontext: As far as I remember the idea was not to update these files while the migration from Transifex to Weblate is still in progress.

ok, makes sense.

yvanzo: what's the state here? could we start using weblate soon?

I can't remember a specific example, but it's possible. However they would have been minor things, and I don't think the tradeoff is worth it. And mind you, I waited for a couple of years before I came to that conclusion, so it hasn't been noticeably useful (while being noticeably a hindrance)

outsidecontext: yes, I was looking into available languages atm

nuke it.

outsidecontext: we have a short list of supported languages in MB Server, but there isn't such list in MB Picard.

yvanzo: cool. Very much looking forward to use this for Picard. I had already disabled updates for the secondary translation components in Transifex

The new testing library is predicated on testing inputs and results from the user's point of view rather than testing the implementation details, so i think the move will do a better job at ensuring good UX than the snapshots are doing (because we tend to ignore them as gaff/ unable to review)

Will nuke poc a poc as we rewrite the test files

yvanzo: supported languages for Picard has multiple answers :)

yes, I see it depends also on installers

That should probably be detailed in https://wiki.musicbrainz.org/MusicBrainz_Picard...

yvanzo: for the main UI there is a list of languages supported explicitly in the UI. It includes those languages that had been fairly good coverage. See https://github.com/metabrainz/picard/blob/maste... . Other minor languages with existing partial translations can still be used though, if system language is being used.

Installer and app store metadata are a different thing and only those languages that are complete (or at least nearly complete) are actually used

lucifer: I started using the new popularity datasets and LB radio is actually a bit degraded right now.

yvanzo: not sure why we should list the languages in the wiki. the available languages can be seen in weblate. the list can change, but any translation is welcome

the current popularity set is all MLHD+ and then old dataset was all LB listens.

outsidecontext: the variations en_* have very few translated messages (~1%) for example

outsidecontext: not listing the languages

how much effort would it be to make popularity data based on MHLD and LB and combine them?

yvanzo: ah, got it. misunderstood this.

the en variations are a bit special, because IvanDobsky wanted to edit those parts that are actually different instead of copying all strings.

outsidecontext: just that some components are very independent that it can be worth fully translating some of these even the others are not.

outsidecontext: Weblate has some features for country specific locales.