#metabrainz

lucifer: Wrong month 😝 - https://github.com/metabrainz/listenbrainz-serv...

I would change it in github but I’m not sure if that breaks anything? v-2024-05-05.0 > v-2024-06-05.0. If it’s all good I’ll just do it next time

<aerozol[m]> "lucifer: Wrong month 😝 - https:/..." <- lucifer finally finished the time travel feature

I told people to be careful!! If you make the ticket then lucifer will build it!

aerozol[m]: alright, I guess now we can have it tomorrow.

akshaaatt: aerozol could u guys lemme know when u r free, need to have a meet re mvp and order of priority as aerozol pointed out

Thanks for the chats pranav and theflash_! These projects are going to be fun 😀

yay⚡

<aerozol[m]> "lucifer: Wrong month 😝 - https:/..." <- 😓😓. i'll fix it.

Thanks! We have a few wonky links on our socials now, but it will add a air of mystery and adventure!

theflash__ (IRC): akshaaatt monkey : I have signed off the user dashboard mockups for the iOS app project. They follow the web version closely (yay) so not much to say, just a few housekeeping notes: https://www.figma.com/design/iKmkwUSd4xhwzgsSyd...

Except for Playlists and Created for You, which will have some extra pages/elements not on the figma, it will be good to see the mockups for those when they come up.

Sure! aerozol[m] : Thanks a lot for looking into it :D

monkey: If you catch a minute, would you mind giving a second opinion on the mockups for the LB app user dashboard as well? I *think* I’m as far as I am going to get with it: https://www.figma.com/design/tFLTQiq7QlA1oGRTJV...

I keep going between having those shaded sections gradients, or removing them and making the titles bigger… but I’m hoping that changing the style for all sections across the app would be possible, in future, changing the code once? pranav ?

Ig yes, but we should confirm from jasje (IRC) too just in case..

Hmm cool, well if monkey doesn’t hate the look of the sections we will go ahead anyway

monkey: This is what I mean re. the different headings. The move to gradients and smaller headings came from necessity (imo) when creating the local player/library layout, where it was tricky to tidily navigate a lot of different ‘sections’

local player/library design ^

Let me know your feelings when you have a minute, dr. monkey

Looks nice actually.

What is this? An music player? A scrobbler?

Current GSoC projects to create Android and iOS versions of the ListenBrainz dashboard: https://listenbrainz.org/user/aerozol/

In a general sense LB is a scrobbler, yup!

So, in the long run a bit of a last.fm replacement? :D

I try to get the strawberry music player to recreate something like "smart playlist" from the Guayadeque player that uses the last.fm track.getSimilar from the API :D

Already the last.fm replacement, in my humble opinion (and as a long time lfm user!) And always getting better. Contributors always welcome by the way! I feel like there’s some fun stuff that can be expanded (e.g. in the LB Explore section)

mayhem here is the one working on smart playlists, and getting smarter by the day (which one is getting smarter I’ll leave for you to decide)

Oh, you’re looking at creating playlists from local libraries. That is also a pet project of mayhem’s :)

Hi wileyfoxyx, you can attach your screenshot to a new ticket: https://tickets.metabrainz.org/secure/CreateIss...

<aerozol[m]> "Oh, you’re looking at creating..." <- Oooouh, that sounds amazing. Any more information anywhere mayhem ?

[08:29] aerozol[m] Already the last.fm replacement, in my humble opinion (and as a long time lfm user!)

I'd agree, but imho it's still missing several scrobbling clients:d (I still need one to go directly from itunes on the old mac to lb rather than via lfm)

I dual scrobble everything to lfm anyway, but I have started browsing LB more/checking my stats there :0

yea, that's always fun

just. wish it was scrobbled directly, so i didn't have to keep importing, + I'm sure there are better abilities directly

❤️

especially now that the "import" button is missing and hidden in the settings 😩

natrius: https://github.com/metabrainz/troi-recommendati... can handle content resolving and playlist generation for local music collections, also has some subsonic support (for clients supporting opensubsonic and exposing MBIDs)

.. wait wtf?? I did an import (first one on this machine) and it seemed to go well but tnhe i get an error on a blank page

"You have exceeded your rate limit. See the X-RateLimit-* response headers for more information on your current rate limit."

:D

You have listened to

1,873,788 songs

outsidecontext: Thanks, i added a link to that to the Strawberry Feature-Request.

mmh, we should avoid emoji reactions, especially when directed at IRC users

Its something the people reacting need to be aware of - IRC does not see reactions.

lucifer: reosarevok I’ve been meaning to do it all day, I’m finally making headway… drafting a blog post (without technical bits filled in yet obviously), but I need to update the IRC wiki page as well, so it all works. Did you already have something in mind for the wiki page? I was thinking about redirecting the /IRC page to a new /ChatBrainz page?

MonkeyPython: 🤘

FWIW I know reactions don’t cross the bridge, but I might still do them if it’s for fun (not to communicate anything)

Well, not to communicate anything work writing a line for, anyway

Yes, I agree. It's just bad if the person being reacted to is excluded from that fun part :/

'do you see this 🤘

Sure. We just made a reaction to your post

that's odd i see it when i write in the box but not in the window...

ah that's a font issue

uh... I just see peopel posting smilies liek normal

[09:47] outsidecontext[m MonkeyPython: 🤘

zas: is there a dashboard for cpu usage containerwise on a particular server?

oh!

thanks outsidecontext[m

MonkeyPython: Yes, normal posts with emojis we all see in both directions. But on Matrix you also have this short reactions, similar to e.g. Github or the likes on our discourse forum. And those don't transfer to IRC

It's tempting to use, but we need to keep in mind that this is a chat now shared by different networks and use it responsibly

<aerozol[m]> "lucifer: reosarevok I’ve been..." <- IRC will remain available as bridged, so probably a big banner pointing to a new /Matrix page would do.

lucifer[m]: yes, but this container was never whitelisted, so we don't collect information about it

I mean, a chatbrainz page should certainly mention the bridge, and possibly the Discord too

zas (IRC): i see, according to docker stats. timescale container is using 3-4x cpu than typesense.

New pages would make more sense if we weren’t duplicating all the info about the channels and BrainzBot etc across all of them. Not much really changes anyway - IRC is just no longer the focus. In any case, I will draft a new page and we don’t have to redirect (yet?)

mayhem Thank you for the playlists. Looks good. How have you solved the subevent lookup?

lucifer[m]: ^^

since I restarted the container, it seems back to normal level

zas: nothing weird in logs still. and nothing useful in typesense forums either. i am thinking we should at least upgrade to the latest version of typesense. will discuss with mayhem

lucifer[m]: if you want we collect stats of this container (and if the name isn't meant to change), you can add it to https://github.com/metabrainz/metabrainz-ansibl...

aerozol: I would probably just have one page, mention matrix as the main thing but still mention you can also access the main chat channels through IRC or Discord if you prefer :)

we have a list because else telegraf collects metrics for all containers including ephemeral ones, and it wastes a lot of resources

makes sense, i'll add it later today. container name will probably change during upgrade.

Moin! What are the symptoms of the type sense server? Increased load that never drops, but does after restart?

mayhem[m]: moooin

the load increased since ~2 days, oscillating between 10 and 40, after a restart it dropped for a while. I can be due to nature of traffic too, right?

s/I/It/

Perhaps that's "normal" and we just host those programs on a separate and/or faster machine

Yes, I think typesense is overloaded and. Which is why I am really skeptical about exposing typesense on our API.

So, at this point, we need to know if it's an expected behavior or not, and if there are ways to smooth the load over the time or not. If that's expected, we can throw more resources to it, gaga is used for LB only. How does that scale? can we spread those over multiple machines?

Maybe we should have two servers, one internally for our mapping and one for external use.

Still, not a great solution. Typesense doesn't scale well enough for us

Do you mean anyone can use it?

Yes, the metadata lookup, including the bulk lookup endpoint released yesterday has access to it, if I am not mistaken.

Includes

longer term, we should find a better way to replace typesense. we dont need all the features typesense has.

mayhem’s brain has to juggle a lot of difficult questions

Well, pre-coffee questions are always difficult.

<aerozol[m]> "New pages would make more..." <- The info shouldn’t be duplicated indeed. All the detailed information about IRC is probably better off your new page, that’s what I meant. Thanks for having thought to update this page :)

busy.

<aerozol[m]> "Let me know your feelings when..." <- Hi! The mockups look great! Regarding the headings, IMO the gradients don't look as good and the larger text works better. So for me the cleaner bigger text headings is a-ok

Thanks monkey I will look at mocking up that style tomorrow

Lucifer: i wonder if the canonical recording data, the cleaned/condensed version, could fit into ram

If it does, we can use mnslib to replace typesense.

mayhem: will have to check but my hunch is no.

Let me try it on wolf after coffee and see

I bet I could do it if I use C and manage the memory myself

sure but would it be as stable as typesense? which seems to only malfunction like 1-2 times in a year.

lucifer: reosarevok : I have drafted a blog post re. Chatbrainz. See if you want to add any interesting technical details, or I got anything wrong. You can post it if you like, or give me the okay and I will post it tomorrow. You will see that it links to the new ChatBrainz wiki page (I have not redirect the IRC page yet) - https://wordpress.com/post/blog.metabrainz.org/...

ChatBrainz wiki page: https://wiki.musicbrainz.org/Communication/Chat...

lucifer[m]: The pattern we are seeing is classic overload, which is only going to get worse over time. This is the exact failure mode of our old search server.

i think we should move to SOLR then

We need at least a plan for what to do, should we become fully convinced.

I don't think solar is the answer here.

I see fuzzy matching as implemented by solar/typesense not scalable enough

i think search is a complex problem enough that any solution we come up with will run into problems sooner or later.

*any custom solution

[musicbrainz-server] 14reosarevok merged pull request #3280 (03master…update-deps-2024-05): Update deps (2024-05) https://github.com/metabrainz/musicbrainz-serve...

This is not "search" it is a very specialized form of lookup which is a much more narrow class of problem.