#metabrainz

reosarevok: https://community.metabrainz.org/t/auto-editor-... 🤩

aerozol: ok, mails sending :)

aerozol: re https://docs.google.com/document/d/1aPeAM0wQyPW... you write about adding more columns but honestly i couldn't get it to work well. the pages are already way too narrow (why does an electronic document have wide margins anyway?) and i can't change it or anything

aerozol: there are quite a lot of "aerozol to create tickets" in https://docs.google.com/document/d/1bz32duq95jS... so I assume this is not for me to do (but if you want I can create those tickets and link them in the doc for you to edit. would they be MBS or DESIGN?)

aerozol: I updated the code signing notes

bitmap: any idea whether https://sentry.metabrainz.org/organizations/met... is a userscript or a legit release-editor error?

(I always find it near impossible to figure out the common-chunks errors)

We get a surprisingly high amount of "The “X” relationship between “Foo” and “Bar” already exists." errors

Does that show even if the relationship editor itself detects and blocks the rels?

Or are these dupes getting through to perl somehow?

I've seen this kind of error in the relationship editor.

It probably should not be logged in sentry as it is a recoverable user error.

Oh, wait no, it shows "This relationship already exists.", sorry :)

yvanzo, reosarevok: weblate 5.1 is out and allows customization of the translation workflow per language. That means we have the ability to e.g. introduce a translate -> review worklfow for languages where there is a large enough translator team (and interest to work this way)

it's not yet on our instance, but will be updated in the next days usually.

reosarevok: I think it's legit, I'll check once I'm awke and caffeinated

outsidecontext: sounds good :) For now I think most of our "teams" are currently one-person affairs, but it would be nice to expand that...

reosarevok: yes, that's also why I argued for this not being needed. actually I think we probably first need to build up something like "teams", that also use e.g. the forums to coordinate.

But if there are 1-3 users that actively translate a language and that then also act as reviewers I think this would work well.

bitmap: also https://sentry.metabrainz.org/organizations/met... - any idea if that's expected behavior?

reosarevok: where do you see the high number of relationship-already-exists errors? I only see one with two users and another with one user

but I believe it's possible to trigger this if the relationship gets added in another tab/session, whether by the same user or a different one

I'm not actually sure how to reproduce the first release-editor one, but the "loop" bindingHandler is known to be buggy

the url one is kinda expected due to the way the function is written...there's a race condition where a concurrent transaction can use the url before the DELETE obtains a lock

this is tricky to fix but I'll open a ticket

basically delete_unused_urls needs to use `SELECT ... FOR UPDATE`, and whichever transaction is inserting the conflicting row should use `SELECT ... FOR SHARE` ... I think

bitmap: I see a few every time I check sentry, so I guess it's a few a week?

ah probably my search query isn't finding them

Maybe I'm remembering it wrong, also possible :D

reosarevok: due to flight delay i can make the meeting. My summary: summit.

Ok

All the things I can share, lucifer also knows about, so I'm sure he'll share

"mayhem says: just the right amount of chocolate"

I'm hoping to work half days this week, that is it for me.

It was. The table got cleared up at the end

Hmm. MBS-13296 seems like a tricky one to display

MBS-13296: Whitespace changes not visible in edit https://tickets.metabrainz.org/browse/MBS-13296

bitmap, yvanzo, aerozol: suggestions welcome

bitmap: also, is STYLE-2391 a schema change?

STYLE-2391: add a language attribute to the "Voice of" relationship https://tickets.metabrainz.org/browse/STYLE-2391

(It seems like it'd require us to have languages as attributes the same way we have instruments so I guess that's at least a few triggers/functions)

well, I suppose we could store them as text attributes and validate it on the server until a better solution is available

I mean, I don't think it's the kind of issue where there's a huge rush

But I guess I could make a schema change ticket and put it under the next schema change, and work on it in the meantime

(when I am bored otherwise)

wouldn't it be redundant with the work's language?

Not a work attribute, a relationship attribute

Requests so far using it are STYLE-2391 for artist-artist "voice of character" and STYLE-1162 for artist-release "liner notes author/translator"

STYLE-2391: add a language attribute to the "Voice of" relationship https://tickets.metabrainz.org/browse/STYLE-2391

STYLE-1162: Language attribute on (translated) liner notes relationship https://tickets.metabrainz.org/browse/STYLE-1162

iirc the work languages are just a direct fk to the language table whereas this would require mirroring the languages to link_attribute_type

I finally uploaded my summit photos to the Google drive link aerozol had shared

Yeah, but that's no different than instrument vs instrument link_attribute_types, right? :)

So most of the code would be similar I expect

right, it's just that tha triggers for it are annoying

I can imagine

Luckily we at least don't edit languages very often

[bookbrainz-site] 14kellnerd opened pull request #1027 (03master…bb-data-ts-migration): TypeScript migration of the ORM https://github.com/metabrainz/bookbrainz-site/p...

reosarevok: yvanzo: given that the RemoveEmpty script works on URLs I'm not sure we really need delete_unused_url and associated triggers?

Possibly not? The main benefit I guess is that they are gone immediately rather than after a day, but

That's mostly just relevant to spam I guess

that's the only benefit I can see, but doesn't seem worth maintaining much to me

I would not be too bothered if we dropped it, yes

I mean, the RemoveEmpty script probably has a race condition too, but it's fine if that crashes; if delete_unused_url crashes it brings the whole transaction with it (probably applying some innocuous edit)

Does the RemoveEmpty script restart if crashed?

no but we could at least add some exception handling

the question was: wouldn't [this added relationship attribute] be redundant with the work’s language [attribute]?

And I'm confused how that could be redundant given it's not connected to works in anyway :)

*any way

To follow the example provided in the ticket, one can add a work and set the language to English.

But the relationship is artist-artist, voice actor to character

Maybe more examples are needed to explain how it can be needed.

Well, all the Disney films for example will have different actors voicing the (singing) characters in Spanish and in English :)

Ok, IIUC, it’s the whole relationship “voice of” that would be redundant with artist-recording relationships (I expected but doesn’t seem to exist) on a period of time.

(There is no relationship to say that an artist A interpreted a character C on a recording R.)

uhm

Yeah, we cannot have that right now because that would be a 3-point relationship :)

bang?

We can only say artist A sang or whatever

MonkeyPython: in one min, from my clock!

:)

oh, it's 19:03 here

<BANG BANG>

my baby shot me down

<https://www.youtube.com/watch?v=2ips2mM7Zqw>;

Welcome to the first post-summit MeB meeting!

Hope you're ready for more of the good old talking to your screen rather than to physical people

I have two written reviews:

mayhem didn't send anything specific, but basically said "GSoC summit" and "we ate all the chocolate"

Living the life

And aerozol said:

lol

'''

I had something big on earlier this month… hmm what was it… nope, can’t remember.

Anyway, since the last meeting I:

Made some tiny PRs - a small step for devs, a giant leap for aerozol-kind - to add new players to the ListenBrainz-enabled apps list, which is looking nice and healthy now (https://listenbrainz.org/add-data/)

Did some work on a nice MeB OBS/streaming overlay that would probably not work and we would end up having to use zoom instead (huh, that’s a weird thing to imagine)

Spent a ton of time on a plane (hmm, why was I on a plane??) sanitizing and trying to make sense of 2017 survey data

Some interesting mockups and discussion on MBS-13292 + MBS-11748 (to be continued)

MBS-13292: Display partial translations in language drop down https://tickets.metabrainz.org/browse/MBS-13292

MBS-11748: Releases are not ordered by date when showing all releases groups https://tickets.metabrainz.org/browse/MBS-11748

Spent a whole weekend writing notes and drafting an enormous blog post on something big that happened early this month… strange… nope still not ringing any bells

The person with the best memory can go next