#metabrainz

jivte: hi!

Hey!!

Can you please review my detailed album page design :)

jivte: Looks like a great start! I will leave some comments

aerozol: Thanks comments improve design a lot (learning from add listen design) :)

jivte: is there a ticket for this? I thought there was, but I can’t find it

jivte: I think this is looking really nice! Everything is in my comments, but overall I think it looks good, and a very good start that we could add more complexity to later.

I’d be keen for monkey to jump in and have a look too, once he’s up!

aerozol: No there was no ticket

actually i was planning this feature for gsoc so just needed a little help from you anyway thanks

was asked to develop the design myself

:|

jivte: All good, good luck with the submission! I’m sure we will work together more if you get it :)

yeah will show design in today/tomorrow meet and yes that comment on addon features will discuss with monkey

jasje I had fixed that issue already from what I remember

Are you on the latest main?

lucifer: is this text still accurate? “Importing the same listens from two different sources such as Last.FM and Spotify may cause the creation of duplicates in your listen history. If you opt into our automatic Spotify import, you may notice duplications in the last 50 listens on Spotify.This is a temporary issue while we find better ways to deduplicate listens.”

akshaaatt: should we link to the listenbrainz app here, under mobile devices? https://listenbrainz.org/add-data/

I would also like a prominent mention on the homepage aerozol

Usually bigger websites have it on their footers

WOULD YOU NOW

In the footer sounds good, since we’re wanting to move away from people going to the homepage much

Right

It might be worth doing a ticket for that akshaaatt, I’ve made a note in my mockup, but there’s so many things to be implemented there. Unless you’re happy to wait

Sure

Will add

Hmm what license is all the ListenBrainz data made available under? I can’t find it anywhere

Not on the LB website or in the docs. Is my search-fu failing me once again?

hello, I am Kapil, a 2nd year Computer Science undergraduate . I am new to open source contributions, but I have good knowledge of C, Html, CSS, JS, React and NodeJs . I would love to contribute to your organisation . Can you suggest me some Good first issues ?

akshaaatt: i synced but the updates still didn’t come (Git desktops fault)

anyways looks like the issue is fixed ill revert my commit :P

That’s fine jasje_ but let’s avoid raising a concern until we are absolutely sure about something

Otherwise it would end up leaving a false impression of the fixes we’ve made

:P mb didn’t see that merged pr

actually i got little time left yesterday and remembered the convo some days back

It’s fine jasje_ :) let’s just take care in future

akshaaatt: i had some questions

Anything urgent?

well no

go go

Awesome then

River rafting!! have fun akshaaatt

Thanks a lot aerozol !

aerozol: access granted!

jasje: thanks! Now - what do you want me to do :)

The upper section looks bland to me like

the username pill isn't just feeling right

neither the follow following btn

also, not to mention the interface for a user to view someone else's profile

where to fit the follow button

the main problem with the design is that we can't include the profile picture which cancels out all the design i see online :(

jasje: Cool, I’ve added it to my notes on the ticket: MOBILE-107

MOBILE-107: Profile page https://tickets.metabrainz.org/browse/MOBILE-107

Noicee

Cause you’ve caught me at the end of my workday. I have a few urgent things to do for monkey that he asked for last week, then I’ll dig into your mockup :)

aerozol: no problemo, ill be fixing my design until then!

💪

monkey: I’ve completed all the settings and about mockups. I’m feeling good about it! Let me know your thoughts: https://www.figma.com/file/YRbCOtFHBez8XmMdCKbG...

(at the bottom)

monkey: I also did a mockup for how we could identify the user more clearly in the top menu. I didn’t workshop this much, but thought I’d check in if this is what you’re imagining. Or if you already have a great idea https://www.figma.com/file/YRbCOtFHBez8XmMdCKbG...

Tomorrow: How the sub-navbar works on mobile, and then I think I’ve covered the things we discussed Friday?

jasje: Currently a track can be pinned multiple times but I wanted to propose a solution for pinning a track only once. Can I raise an issue for that?

shelly: im probably not the right guy for this but as general advice you can always make tickets if you feel something is wrong and eventually youll have a response if the design is intended or not :)

ping mr_monkey maybe

jasje: Sure thing.

O’Moin

moin!

For some reason, I again can't get to JIRA

It works for everyone else, I assume?

it works for me

Thanks zas, I checked logs, looks like an HW failure indeed. I’m not sure why PG refused to start either; bitmap might have a better insight of the “pooler error”.

mooin!

yes, it looks like a pure HW reset, nothing got reported by the kernel, and nothing was unusual before the crash

PG didn't close cleanly in this context, so perhaps that's why it failed to come back

Sigh. Well, phone hotspot it is for now :)

reosarevok: Added a ticket for the sluggish report.

Thanks

aerozol: yeah that description sounds right.

mayhem: about the troi rate limit issues, i think directly accessing the database is the best solution.

i tried using the extended ratelimit with approved token but that exhausts soon too.

would likely work for a few days and break again if more users register for daily jams.

remember that we can change rate limits on a per key basis.

we can make troi's key unlimited.

should we try that first?

because instead of direct DB access, I would much prefer to port solid patches to spark.

sure can try that.

yeah i guess if we port it to spark then adding db access isn't worthwhile.

porting to spark has always been the plan -- troi is literally just the sandbox to learn how to do it. but weekly jams and weekly new jams are likely the first candidates to be ported.

once those are shaken out.

I wonder about the best way for troi to ensure it has a high rate limit since those are kept in redis.

might just be good to have a small function automatically set the key limit to something huge before we invoke troi for running recs.

what do you think?

mayhem: we have define those on app init, currently it is 50. we can just change it to say 50000.

yep, that

but lets not add that rate limiting code to troi -- it think it should live in the spark code that calls troi

uh not that ratelimit code will go to LB main webserver.

when do you plan to call the rate limit setting code? at start of app? what is redis is reset after the start of app?

mayhem: the ratelimits are configured in python. the current number of requests in a rolling window is in redis. https://github.com/metabrainz/listenbrainz-serv...

the current number of requests is in redis, the threshold to check it against in python and we already have a function to configure the threshold in BU.

but doesn't that chance the rate limit for *all* tokens? that's not what we want, is it?

yeah that changes the ratelimit for all tokens but we only have one token currently.

every user has a token.

the per token validation has this validation function to check which tokens are valid.

in LB we define this function as https://github.com/metabrainz/listenbrainz-serv...

so afaiu, the extended ratelimit only applies to the troi tokens.

huh, the semantics of what I had written back in the day seems to have changed then.

my intent was that anyone using an auth token gets a higher rate limit. like 2x over the normal limit.

but now it seems that token use is really limited to special tokens, not all user tokens.

does that seem right?

right.

to allow it for all users, we'll have to hook up it with LB user table.

ok, then your approach should work fine.

looks good to me. thjanks. :)

and thanks for explaining how the ratelimiting changed.

cool, let me do a test. and generate daily jams for all users across timezoes.

smrt

lucifer: the RecentListensTimestampLookup patch in troi is becoming more important -- pretty much for all troi patch use cases.

but it is rather ineffocient, so I think we need to build yet another freaking dataset. :(

that one can be quite well optimized on spark

exactly that.

so lets move the daily jams to spark?

the basic building blocks will remain same i guess.