#metabrainz

put the "2023-01-18T14:28:13+05:30" representation in the tooltip

yes, the tooltip does show that

mmm, of course - I see the reason to use the offset value in the "standard" case of daylight savings

because it'd be nice for it to show 4pm when I was actually listening to it at 4pm, not 3 or 5

monkey: hi! have you used luxon before?

I don't want to sound like I'm saying "stop everything and throw it away", though I can see how this might sound like that

but I wonder if we should stop, make a ticket, and list all of these cases

and get aerozol's feedback

and work out what we want to show in each of these cases

I can think of a few cases off the top of my head: 1) daylight savings, 2) someone gets on a plane and goes somewhere else in a different tz

Sounds sensible to me

after thinking through it, I think you're right in the way that you've shown the data in your screenshot

My alternative proposal: use browser time for display on the list, show submitted time + timezone on hover or with some see more option

ii agree but i don't think we'll figure out the cases until we do put it for testing or regular use.

it's just that this is a very uncommon situation to be in. normally you're going to submit all in one tz, and then there'll be a switch and it'll all be in the other timezone

reosarevok: right, I initially thought that (and that's basically what we do now), but that breaks daylight savings

as in the summer, all of the things you listened to in the winter will be an hour out (ok, it's a small thing, but this idea of introducing the timezone was our way of preventing this from happening)

reosarevok: i think if the datetime iso string include the correct dst offset then dst would be observed fine.

uh that was for alastairp ^

I'm wondering if we could have a little "/!\ the tz in this listen doesn't really match where you are, but we're showing it as submitted" popup

because my hunch is that tz-data accounts for this.

lucifer: yes right, I think we're both agreeing on the same thing

ah ok, makes sense.

there's also this thing of how we show 3 hours ago, 4 hours ago thingy.

(and this is kind of why I'm asking for a ticket - for me I find it much easier to look at a list of our decisions instead of always trying to remember what we decided in the last discussion)

like aerozol listens a track at 11 PM NZ time, when i look at his page, should it show as 6 hours from now or 1 hour ago?

1 hour ago

makes sense, will have to account for that in tz handling.

reosarevok: O

I've been thinking about timezones :)

haha

Ok

just looking at the spec again

LB-1213

LB-1213: How to display timezone aware listens to users? https://tickets.metabrainz.org/browse/LB-1213

<3 lucifer

i'll open a draft PR for now because except test/docs its mostly done.

> The Schema Object allows the definition of input and output data types. These types can be objects, but also primitives and arrays. This object is an extended subset of the JSON Schema Specification Wright Draft 00.

reosarevok: ^

Hmm. So maybe anyOf and allOf combinations would work?

anyOf for the incs, and allOf inside for incs that depend on other incs?

anyOf [aliases, recording-rels, allOf [recording-rels, isrcs] ] sort of thing?

yeah, right. I'm looking at anyOf at the moment and seeing how it gets represented

akshaaatt: apparently the first startup bug was not fixed. The bug is inconsistent and might have not occurred while I was testing my fix.

zas: DNS issue is resolved then, yes?

yes, it was one hour after the issue appeared.

ok, cool.

reosarevok: I've got it generating the interface and it'll construct a valid query. still trying to set the valid parameters

oh wait, it's not a schema

it's an enum

sigh

reosarevok: it seems that "you can only include x if y is set" is going to include a twisty maze of anyOf/allOf

I think so, yes

but is that strictly necessary? e.g.

> If you request work-level-rels for a recording, you will still need to request work-rels (to get the relationship from the recording to the work in the first place) and any other relationship types you want to see

(from the wiki)

would could just add this as a description to the specification on /release

at least initially

I guess

We will return a bad request in that case, with the error

$i is not a valid option for the inc parameter for the $resource resource unless you specify one of the following other inc parameters: $params

But I guess if it's documented maybe it's fine as a start

and the inline "try it out" does nicely show these error messages

That's their actual suggestion, it seems, so :)

OK, let me finish up the artist example and post it on that ticket

oh, I wonder how to separate browse and search into 2 separate docs, as the URL is the same, right?

Sigh. Yeah

That's kinda the "mutually exclusive" thing there

It's one or the other depending on the params

It's basically "if there's ?query it's a search"

(and documenting ?query itself might be *fun*)

description: "A lucene search query"

done

:)

I mean the actual parameters that can be part of a query

But maybe it's not hard, dunno

Anyway, I guess yvanzo is waiting for a second meeting? :) So if you want to put what you have up, I can look at it a bit

(are you playing with it in https://editor.swagger.io then?)

reosarevok: ^ paste that into editor.swagger.io if you want

Ok, I could play with that while you SOLR

hi yvanzo, can we talk starting in ~20 minutes? Or after lunch (approx 2pm CET) if that's better

monkey: Hey for that throttle change it is not for promises and searchUsers is a promise based utility so should I use throttle-promise library or something else can un please help in it. :)

alastairp: as you prefer :)

yvanzo: after lunch, in 1h45m would be much better for me, thanks

Ah, good old Spanish lunch

👍 🧑‍🍳

reosarevok: we have people in the lab who don't eat with us at 1:30, because that's too early for them

I meant more the "lunch taking over an hour" bit :D

hey, I didn't say that I'm going to lunch now!

But yeah, my family ate fairly consistently at 3 PM actually

right, you started a few hours ago?

Maaaaybe 2:30 if we were in a hurry

🙄

alastairp: lol, so I guess *technically* we can make it so Swagger is happy with splitting browse and search by having /artist: and /artist/:

🙄 indeed

Not sure if that'd even work though when trying to use try it out

Getting really tempted to add an alias mbid/search?query :p

seems like it might be https://stackoverflow.com/questions/40495880/ho...

> Basically, you can have two definitions to the same path by adding a slash (/) in the URL.

:D

I have to say, writing docs is pretty rewarding when that is the only goal for the week.

This also means that it is impossible to have multiple paths that differ only in query string, such as: [what we want to do]

This is because OpenAPI considers a unique operation as a combination of a path and the HTTP method, and additional parameters do not make the operation unique. Instead, you should use unique paths such as: [what reosarevok wants to do]

Yeah. tbh our API design is kind of a hot mess

from https://swagger.io/docs/specification/paths-and...

So I'm not surprised it causes issues

But better to document it semi-properly even if it's a bit hacky now, then improve on it

So if //artist or whatever works, then I'll try it out

tbh, generating 2 sections with an extra / or ? is probably a great temporary solution with a small note

as long as MBS supports the auto-generated URLs

Yeah

I'll go for a walk to leave the house before it's dark

Then come back and play with that idea

(and maybe comment on whether I can follow your SOLR docs at all :p )

lucifer: I saw your message yesterday and wondered "WTF is a Luxon", so that should answer your question :D

Gotta say, luxon looks nice to handle timezone stuff

Also I like the sass in their DST docs:

>Because our ancestors were morons, they opted for a system wherein many governments shift around the local time twice a year for no good reason. And it's not like they do it in a neat, coordinated fashion. No, they do it whimsically, varying the shifts' timing from country to country (or region to region!) and from year to year. And of course, they do it the opposite way south of the Equator. This is all a tremendous waste of

everyone's energy and, er, time, but it is how the world works and a date and time library has to deal with it.

Oh, thank fuck someone solved this "where do I put my modal in my react code?" issue for us: https://github.com/eBay/nice-modal-react

yvanzo: sorry, it took longer than expected to get my lunch, I'll be another 15mins

np, take your time :)

[acousticbrainz-server] 14kenrick95 opened pull request #428 (03master…fix-broken-link): Fix broken link to Metabrainz https://github.com/metabrainz/acousticbrainz-se...

zas: Would you recommend any list or API for scanning URLs?

(so far urlscan.io and virustotal.com would be quite pricey)

[acousticbrainz-server] 14alastair merged pull request #428 (03master…fix-broken-link): Fix broken link to Metabrainz https://github.com/metabrainz/acousticbrainz-se...

yvanzo: hi, I'm here now

yvanzo: nope, but what's the purpose?

zas: scanning URL edits

To find what? Spam? Broken URLs?

using wolf atm for debugging mb solr session

yvanzo: I'm running `docker-compose logs -f --tail 100 search`

yvanzo: but http://localhost:8983/solr/#/recording/core-ove..., pointing to wolf shows content

yvanzo: if that's about identifying unsafe websites, maybe https://developers.google.com/safe-browsing/v4

yvanzo: when you get a timeout, I see this error message on solr

The request took too long to iterate over terms. Timeout: timeoutAt: 11426689695006838 (System.nanoTime(): 11426689945661048), TermsEnum=org.apache.lucene.codecs.blocktree.SegmentTermsEnum@2d9515ef

[23:08] <aerozol> Just rediscovered this snap I took from a co-workers desk years back. Too good! https://usercontent.irccloud-cdn.com/file/itNZg...

<-- me when i play games :D