#metabrainz

hello

did not know metabrainz had a a bridge for matrix

can someone help with this one? I integrated external lib 'react-markdown' but its not failing at tests... (full message at <https://matrix.moviebrainz.org/_matrix/media/v3...>)

I tried to change jest.config, still confused how it works

Someone is asking at support "How high is traffic to your website? How many people access it daily?" - do we have an estimation I can give them?

(not sure why they care but)

Now I want to know too! Otherwise I know there were a bunch of impressive MB stats in the summit reports (definitely ’22, maybe ’23?)

The blog post for 2022 only says ’150 Terabytes served this year’, but I know the slides had some more details. Courtesy of zas

O’Moin

bitmap: Not too odd given the recent maintenance operations plus the cache change, but worth watching it.

monkey (IRC): in ur YIM feedback u had pointed out that the text on your axis side is not readable in white.. it wasn’t in white earlier as well.. could you please let me know what you were referring to exactly.. thanks

yvanzo: moin! the broken Dutch strings that are filling our Sentry should go away on Monday, right?

(well, this week it's mostly "stuff was down/broken" but)

reosarevok: if fixed or marked as fuzzy, yes

IIRC you marked them as needs editing, so I guess that'll do it :)

Thanks

yes, fuzzy is implied by needs editing

lucifer (IRC): the image for the yim art from the backend takes a while to load.. that is causing the image to not be fully loaded in the app so can we share a base64 encoded image from the backend and convert it in Frontend? Won’t this be faster

pranavkonidena[m: images are hosted on Internet Archive whose servers are sometimes slow, so the backend also doesn't have direct access to those images.

Ahh i see .. then it isn’t possible

you are using the 250px thumbnail versions right?

Yup

The overview image is not able to load tho while sharing yim on LB Android

yeah i am not sure this can be improved further then

Yeah cool thanks

the overview image only has 2 cover art though so its surprising

so the text loads for you on the image but the cover art images don't?

Hmm yeah.. but even while hitting the api endpoint via browser, I noticed lag for the image to load up

lucifer: Yup exactly

yeah not much that can be done here

reosarevok, cc aerozol: ~30M hits/day to MB website.

[listenbrainz-android] 14pranavkonidena opened pull request #359 (03dev…YIM/MinorFixes): Minor fixes for YIM https://github.com/metabrainz/listenbrainz-andr...

That includes JS calls for editing though.

website + ws or only site?

website + ws/js

Woww 30M damnnn very noice

Actually it seems to include ws/2 (API) too

Ahh… but still very impressive imho

Thanks

yvanzo: I figured out I missed adding context for one 'Time' string, sigh

I'll just push it to master and it can be put on next beta, whatevs

Given that we already have some stuff waiting in `master`, it will likely go to `beta` as soon as Monday :)

Perfect

reosarevok, aerozol: ~1.5M hits/day for the MB website alone (not counting ws/js either)

mooooin!

1.5M for the website is deffo an increase in website traffic over the past.

lucifer (IRC): could you throw one last glance onto https://github.com/metabrainz/troi-recommendati... today? there have been a lot of commits since you approved.

Hi, have a technical question about one of the APIS found at https://datasets.listenbrainz.org/

I had been using https://datasets.listenbrainz.org/popular-recor... for some time here (https://github.com/regorxxx/ListenBrainz-SMP), but just noticed it's now gone

I used it to retrieve popular recordings for a given list of artists from other methods, and then scrap the web for youtube links; I knew it was unstable

it's the lb radio + artist query a replacement?

hi! I was just starting to answer you in #musicbrainz:moviebrainz.org

As far as I have read, it should be on easy mode with :nosim (?)

Oh, just seen it

yep, it's not listed on the stable web

LB radio uses popularity data, so the data is out there. :)

remind me of the old API end point. given artist, return top recordings?

; // [{"[artist_mbid]": "69ec6867-bda0-404b-bac4-338df8d73723"}, ...]

As input

/ [{artist_mbid, count, recording_mbid}, ...]

as output

lol, we need to reorg our API endpoint docs. too many endpoints.

isabelxxx: https://api.listenbrainz.org/1/popularity/top-r...

my browser history never forgets: https://api.listenbrainz.org/1/popularity/top-r...

mayhem[m]: reviewed, lgtm

thanks!

isabelxxx (IRC): sadly, that endpoint only takes one artist at a time.

the good part, its now a stable API endpoint.

this endpoint also has popularity data, but we'll likely change this endpoint soon: https://api.listenbrainz.org/1/lb-radio/tags?ta...

The first one worked like a charm, will have to make multiple calls for arrays, but worked without many changes

going forward, the plan is to include popularity data in the recording metdata cache, which supports fetching multiple items.

I was also going to ask for https://datasets.listenbrainz.org/recording-fro...

but I think that's the second endpoint

yep

yvanzo, outsidecontext: is it possible to disable the Apertium suggestions in Weblate?

They're so amazingly bad for Spanish at least that I would much rather nobody could ever get them :p

Hehe, I've thought that as well about the German suggestions

Where do you see these?

Anyway, now disabled.

In "Automatic suggestions"

Thanks

Lol yvanzo, I was just looking for them again and they were no longer there :P

Nothing was showing up for me under “Automatic suggestions”.

It might be that Apertium doesn’t have English > French.

Or is it just their website that is limited at purpose?

Is '2a01:4f8:c01e:70::1' one of the MusicBrainz servers, or maybe one of the internet archive ones that serves cover art?

Seeing very long connect times to it.

Port 80.

whois says it's a Hetzner hosted server.

moon.metabrainz.org -- that is our load balancer. not under our control. mehinks.

hetzner's load balancer, to be exact

Do LB statistics work solely based on track title and artist name while ignoring anything else? My recent listens are correctly mapped as expected using their recording MBIDs, but statistics look like they mismatch/conflate similar track titles regardless of MBIDs.

mayhem[m]: Oh ok. Something seems to be amiss. Could it be one of the backend servers not responding so it's taking a while for the load balancer to move on to another?

adhawkins: which domain do you access? URL?

Possible

zas: Hard to say, it's my tagger application (flactag) that seems to take a long time. It seems to be getting the WS responses quite quickly (from my local mirror), which is what makes me wonder if it's the coverart responses that are slow.

If it's CAA, most are redirects to Internet Archives servers, but hard to tell if you can't provide more details about queries made

Understood. I've just re-tried it and it seems responsive now, so even if I add some debugging to the app it probably won't help. One sec.

Meh, still quite a few "Series" strings that do not indicate singular or plural

zas: LB is having issues connecting to PG cluster ^

yvanzo: pushing that to master as well, I think? :) https://github.com/reosarevok/musicbrainz-serve... can you quickly verify?

Hi, I'm interested in contributing to the resolution of ticket BB-620 (https://tickets.metabrainz.org/projects/BB/issu...). Could someone please provide guidance on how I should proceed with this issue?

BB-620: Allow measurements in more formats

zas: it seems to have resolved itself. but no idea what caused it

monkey: ^^

it seems that LB open too many connections

which graph is that?

Number of backends connected

ah okay, makes sense

even musicbrainz and all the backends it has do not reach this level