#metabrainz

alastairp: Since the Final submission is editable until 11:30 PM IST today, I TEMPORARILY went ahead and wrote everything to a medium.com blog, and submitted the link.

Rest assured, it'll be editable till today midnight, so we'll have enough time to replace it with a final blog hosted on blog.metabrainz.org instead.

I was about to sleep, but was feeling very uneasy about the creeping deadline. So just to be safe, I just finalized the blog as much as possible and submitted with a medium.com blog since it can be edited anytime. Let's get on with the final review and publishing of the blog whenever you're free today :)

🎉 Happy Diwali! 🎊

Happy Diwali!

hi Pratha-Fish, congratulations on getting your report finished, I got notifications all weekend so I can see that you were working hard

I think it's looking pretty good now, I'll make just a few more small suggestions to tidy up the text

Pratha-Fish: do you have an account on blog.metabrainz.org yet? If not, you should ping Freso as soon as possible to ask him to set it up

Mornin' !

moin!

mayhem: Got an SVG tidbit for you :) To have a non-square image cover the whole area: add `preserveAspectRatio="xMidYMid slice"`

That'll center and crop

ohhh, let me try that.

Will definitely instantly improve the square grids artifacts

I'll try to get you an appropriate transform string for the images in the cover-on-the-floor , with perspective distortion

that snippet doesn't improve things: http://136.243.82.226:8000/coverart/cover-art-o...

oh applied to wrong image. hang on.

ohhh, that did fix it. THANK YOU!

awww, yiss. :)

alastairp: Haha It was the blog was more complicated than I thought. I planned on finishing the blog in 4 hours, but it kept compounding and took me a whole weekend and a full night. 🥲

Pratha-Fish: and now you see why I suggested that you try and get the first draft done by last wednesday?

(and why I've been suggesting it almost weekly for the last month or so)

Yes ⚰

ahhh, yiss, fix applied to all the other images. drastic improvement! thanks monkey

because wow, I can now see all of the diffierent things that we tried! a lot of interesting investigation

Yusss

sometimes these projects just consist of trying something, seeing that it doesn't work, and then moving on to the next idea

I expected the blog to be 2-3 pages, but it turned out to be 23 pages long 💀

Pratha-Fish: in the next hour or so I will give a final set of feedback (I'm sure we can find some things to cut back, or break out into separate posts)

alastairp: yes definitely

also, I don't have a blog.metabrainz account yet!

CC: ^ Freso alastairp

I'm sure you want to spend most of today at Diwali, so let's try and spend the minimal time necessary today to clean it up, and then we can pick up the last coding steps tomorrow/wenesday

Thankyou! Really appreciated :)

lucifer: Hello! I know there have been some reports about MBID mapping and album artists vs. track artists; I stumbled upon a good example of a VA release that ends up appearing many time in the top releases: https://listenbrainz.org/user/jukevox/charts/?p...

In case it is useful for testing

(For the Birds: The Birdsong Project)

And somehow one of them is not mapped: https://usercontent.irccloud-cdn.com/file/PgvRC...

monkey: http://136.243.82.226:8000/coverart/designer/ro...

look at it in the browser and then download the SVG and look in the desktop\

sorry whoever is trying to load the above link, but shit is broken right this second.

is this weird animation expected when deleting listens? https://usercontent.irccloud-cdn.com/file/U27V3...

deleting listens seems to have been removed in beta, so I guess this won't be an issue going forward!

Animation expected, weirdness not so much :)

I suppose we need to review that animation

mayhem: https://usercontent.irccloud-cdn.com/file/8uoa7...

aerozol: https://grid.listenbrainz.org/coverart/designer...

atj_mb: did you read about loki's storage options? If not, can you so we can talk about this later?

with scalable gateways, moving to loki looks like a good idea, it can also provide a solution for web analytics, and give us much better tools to explore logs, especially in containerized envs. So, on paper, Loki's answering to a part of our needs, but I have tons of questions about a valid production setup, they are many options to evaluate.

for those interested -> https://grafana.com/oss/loki/

Pratha-Fish: hi, I've finished adding comments to your post. In general it's _much_ better than the version from last week, thanks for your work on it. The comments are mostly to clean up just a few small things

zas: looks like we'd need an Apache Cassandra installation

$$$

atj_mb: or minIO cluster

in terms of keeping the focus of the blog post more specific, I think that it might be better to skip the list of things that you learned + the list of URLs, I don't think that they bring a lot to the main story of the post, and they take up a large number of pages. this could be a good candidate to break out into a second post

alastairp: Thanks, I'll check out the comments ASAP

I also agree on the list of things + list of URLs part. Let's just leave those out from the final blog

I was pondering about the same issue too. Since this blog is going to be like a landing page for what we did in the past 18 weeks, and a lot of curious people might visit it. I think we need to figure out how not to bore them instantly and highlight the most interesting parts about the story too.

ig I'll move ahead with the issues you highlighted for now

alastairp: I think we should also get started with publishing the blog on blog.metabrainz.org ASAP, considering I am totally unaware of how to publish a blog on that particular site, and might have to spend time asking questions and all

did Freso make you an account? if not, I can do that.

Pratha-Fish: Did you ever e-mail me about getting access? I don’t see anything from you. :\

Freso: Oops I wasn't aware about the emailing part! My apologies

(I thought we'd be discussing that in the IRC channel itself)

Freso: What details should I email you? (and where?)

😂 I _just_ received an e-mail “Can I Post an Article On metabrainz.org?” and I thought it was Pratha-Fish, but no, just another spammer. 🙃

Pratha-Fish: Your Wordpress account e-mail if you have one, or if not, the one you’d want to use. :)

lool

Freso: I don't have a wordpress email account yet. Should I go ahead and make one?

Nah, just let me know the mail you want to use.

alright

mayhem: about the array support so how about we do this? a query can specify array parameters as `[name_of_param]` ([] brackets being the key). the ds hoster will accept params for both `name_of_param` and `[name_of_param]`. automatically changing the first to an array for the query to receive. this part is important for web viewer. for post a direct array for the param and for get comma delimited values to denote an array.

that sounds sensible.

[listenbrainz-server] 14amCap1712 merged pull request #2218 (03master…fix-daily-jams): Daily Jams Fixes https://github.com/metabrainz/listenbrainz-serv...

[data-set-hoster] 14amCap1712 merged pull request #8 (03master…fix-none-bug): Improve rendering of artist credits https://github.com/metabrainz/data-set-hoster/p...

zas: minio looks easier. I assume we want some form of redundancy?

yup

the fact we're limited to 1gbit/s isn't ideal

mayhem: hmm, actually it seems this conflicts with the existing `[]` support in json get.

meh

tricky, no?

Pratha-Fish: It’s not letting me invite right now. :( It’s telling me “Sorry, we couldn't process your invitations. Please try again later.” Maybe mayhem can do it.

yeah :/. we could add another delimiter or syntax but not a fan of that.

Freso: 😬😬😬

whats are your thoughts on completely revamping this? get rid of the syntax: `[{"x": 1}, {"x": 2}]` totally and do only `{"x": [1, 2]}`

Freso: I get the same error.

:|

Pratha-Fish: Do you already have it written up somewhere?

maybe someone else can publish the post on blog for now and later change the author to Pratha-Fish when the invites work again.

Freso: I have it in a google doc as well as posted on a Medium.com blog

lucifer: might be good to start a google doc or something in order to work out this new design.

mayhem: sure. fwiw, i am suggesting to use the same syntax we use everywhere else in LB.

and while we're on changing the format, the new format of data that is being returned is rather quite odd to me.

returning an array of dicts, with the info I really want tucked into element 3 is really weird to me.

yeah, agreed. maybe the json endpoints could get rid of those markup elements?

Freso: we didn't set up Pratha-Fish's account back in the day because he was still working on the project. Thanks for trying to do the invite

only return the datasets in case of json. but for ui return the markups as well.

Freso: do you know if it's possible to change the author of a post in the wp admin?

ah, lucifer just asked that too, thanks lucifer

I agree that this is probably the best solution for now

honestly I find this convo difficult to follow because we're talking about so many things and everything is a bit confusing. a document/gist to sum them all up so we can see proposed versions would be very helpful for me.

makes sense, i'll start a doc.

thx. :)

mayhem: i added threshold support and truncated the existing table. for days = 7, i am thinking to request without threshold. what about days = 30 and days = 90?

Thanks alastairp :)

alastairp: Thanks a ton!

atj_mb: yup, not ideal. There's a possibility to get dedicated servers with 10gb/s NICs and connect them if they are in the same rack, but there are extra costs ofc.

Pratha-Fish: of course, it's not published yet

and you don't have an account to see it

oh right 🤦‍♂️

lucifer: maybe 5 for the threshold?

mayhem: 👍

zas: I could see minio possibly being useful for other projects in the future

the S3 API is pretty pervasive now

I guess we should do a PoC with some VMs?

mayhem: datasets are in. i added bulk lookup and updated the patch as well. currently it passes algorithm with every item but only the algorithm of the first item is used. once we decide the new formats we can make amends here.

perfect.

algorithms available are `session_based_days_7_session_300_threshold_0`, `session_based_days_30_session_300_threshold_5` and `session_based_days_90_session_300_threshold_0`

nice thing is that calculating these sets is blazing fast! ~4 mins for 7 and 30 days. 8 mins for 90 days.

uhh not `session_based_days_90_session_300_threshold_0` but `session_based_days_90_session_300_threshold_5`

nice. I can't wait to get back to that stuff, thanks!

mayhem: https://usercontent.irccloud-cdn.com/file/HY54u...

monkey: when you have time, please review LB#2219,2211

Add a save playlist button to instant playlists: https://github.com/metabrainz/listenbrainz-serv..., Split production image into multiple stages: https://github.com/metabrainz/listenbrainz-serv...

Will do !

monkey: hi, have you used webpack-manifest-plugin recently (the lastest version)

I have it configured in CB, and it's generating invalid manifest files

Haven't looked at that since we set it up in LB IIRC

Invalid how?

specifically, in dev mode (no hash), I'm getting: ` "main.css": "auto/main.css",`

I don't know where the `auto/` is coming from

hm.

this might be it

in fact, it's even documented here: https://webpack.js.org/migrate/5/#run-a-single-...

I would also have a look at the `output` in wbepack config

Yeah, there you go. output.publicPath

See in LB: