#metabrainz

change paths (/var/ftp to /home/ftp)

Mr_Monkey: np. I'm sure I can think of more :p

reosarevok: Yeah, do my work for me, see if I care :p

70% of hits on the mb website yesterday were from (verified) googlebot (1277045 hits) and yet searching for an small unknown band existing in the db, indexed (search with site:musicbrainz.org shows it) doesn't appear at all over 15 pages of results... discogs appears on 5th page, qobuz on 7th, official sites on 1st/2nd pages, but musicbrainz absolutly nowhere over 15 pages. It sucks.

any SEO expert around ? ;)

Mr_Monkey: if you do the programming, I'm happy to do the asking :p

Mr_Monkey: we should do a more general database cleanup, as well as fixing the invalid relationships

Things like deduplicating alias/identifier tables

until yesterday, the site code re-inserted all the aliases for every entity every time it was edited due to a code bug

LordSputnik: Yes, that sounds sane :) Are there already scripts for cleanup operations?

Oof.

Nope, not currently

We should write some and put them in bb-sql/scripts

Yep

Yeah, it was due to the set handling code, which was fixed and moved into bb-data this summer. It's now unit tested to make sure nothing gets duplicated

KassOtsimine, sorry, I didn't realize that

no prob

i was feeling grumpy :)

reosarevok: i just noticed we don't ask for a disambiguation when creating a new RG having the same name of an existing one, as we do for artists. Asking for one would reduce a lot duplicates, is there any suggestion/ticket you're aware of for this ?

We mostly don't enter rgs by hand, but automatically

zas: if on the same artist maybe. but else it could get pretty hairly quickly. other thna that good idea!

i mean we can ask for disambiguation as well as confirmation when a new RG of the same name exists for the same artist

LordSputnik: I'll create a ticket for it so it doesn't disappear. I still suck at SQL and I feel like I would fuck it up without realising. Would you have the bandwidth to take care of those cleanup operations (not the creator deduping or anything like that)?

it might even help against duplicate releases

just created those (artist called releases "EP", perhaps it should be self-titled, but that's the same)

in general, i think we should have better safe checks to prevent common mistakes to happen (like creating a new RG when not needed, which is VERY often)

or check packaging set to None on a physical release, or to anything but None on digital releases

chhavi_: ^^ included in your plan ?

zas: well not setting packaging on physical release ,.. often i don't know waht packaging my dumb library has wrecked so i know not which it is

that's not what i say

oh? i missunderstood

CD packaging = None is VERY rare (we should warn about it, not forbid it)

oh no explisitly *not set* but explisitly set to *none* ah

and digital release with packaging digipak is also uber-rare

now i get

same for capitalization rules, they have to be linked directly from release editor (depending on selected languages), a new editor hardly knows about then.

Mr_Monkey: Yeah, I can do that at some point

Happy to have that assigned to me

Having better infos from the start would reduce a lot edits

:o I thought cap rules where! yes i agree

LordSputnik: Thanks, that would be very helpful. I don't think it's a huge rush as it doesn't currently keep anyone from using the site or database correctly, if I'm not mistaken.

Merging! Merging! There are two Lovecrafts! ;)

I guess there is a ticket for that though

oh no two lovecrafts?! thats like dividing by zero!

rather than merging entities, i'd like to prevent them from the start (whenever it makes sense), it reduces the amount of work for all, and the size of the db....

Both are needed ;)

of course

:))

i rememember *distinctly* saying something like this about recrordings when NGS was happening

But yeah, that too

;)

KassOtsimine: not much to be done with that though was there? Other than risking merging the wrong recordings automatically

That was a mess we couldn't probably solve properly in any way :/

¯\_(ツ)_/¯

LordSputnik, Mr_Monkey: what should happen with introductory notes by editors? New work too?

oh shit that's a thing that's separatr? D:

In the near future, yes, I think so. There isn't currently the correct work type, though.

huh. our "temporary" FreeDB gateway has been in operation for 11 years and is still getting 2 requests per second.

well, enough of that. :)

I will very soon open a big discussion about the BookBrainz schema and how everything fits. I know the discussion has been had before, but I want more :)

we kill the gateway and the next day brexit happens. should be a fun week.

:D

!m ruaok

You're doing good work, ruaok!

Mr_Monkey: yeah, I'd like that too

reosarevok: I'm sure I've seen it suggested somewhere : I wish there was a way to link works with BBIDs ! https://www.youtube.com/watch?v=GHyMCFLQ2EE

Mr_Monkey: you mean MusicBrainz works or?

Ye,s sorry, MBIDs

reosarevok: I'd say make a sub work and link it to the parent work

(for editor notes)

What if there's no parent work :)

(this is an anthology of short stories)

Oh, actually, make a work then link it to the edition

We could have a special relationship for that

Eventually I'd like to have the concept of sections of editions, and the ability to link works to those

buh, dodn't we already have a "link to MB entity" lnik type? or maybe notl ink, yo ucna put oin the mbid in the section for various codes isbn etc

oh halp. i also have an anthology of short stories to add :s

reosarevok: yeah KassOtsimine is right, you can link to MB works in the identifiers section

Sure :) But that only helps if they're the exact same

Mr_Monkey: ^

I think Mr_Monkey's example was more "is the basis for MBID" :D

Isn't that better stored in MB?

Yes, a relationship of type Work inspired Song/Album

Oh right

No, there's no way of saying that at the moment

Technically possible, since relationships are a pair of uuids and a type, but we'd need to do some weird stuff to make it work

LordSputnik: sure, in this order it would, possibly :) Although I'm sure there are some which are the other way around

A good first step is relationships with areas, since BB areas are replicated MB areas

zas: we have not reached editing interfaces, so not in plan yet

reosarevok:

iliekcomputers: when you got a sec, I wanna compare notes on spark/hadoop/yarn, etc.

the spark universe has so many moving parts it is hard to figure out what we need and what we don't need.

The Spark Universe should be your new band, ruaok

LordSputnik, Mr_Monkey: is there any diagram of the current BB schema?

reosarevok: Yes, let me find you the url

Ok :)

Just to have some idea

Thanks!

outsidecontext, zas, samj1912: does one of you know if something has changed in the way picard handles exceptions? I'm pretty sure picard 1 basically caught every single exception, so you could at least save already correctly associated albums etc. this does not seems to be the case anymore with picard 2

"Congratulations! MetaBrainz Foundation has been selected as a Google Code-in 2018 mentoring organization." Woop wooop !

yayy!

Damnit, I guess we need some tasks then!

ok wtf. how am I supposed to select that Neil *wrote* https://bookbrainz.org/edition/ef52527a-f68b-41... ? i tried in both ends. in nei'ls end and in this. i only get every at other than "wrote" what?

CatQuest: That's an Edition you're looking at. The 'wrote' relationship is between a Creator and a Work

Mineo: not AFAIK

I find it confusing as well.

https://bookbrainz.org/creator/0fcd19b7-d57a-40... already has two links to editions though :D

so yea that is *SO* not clear

can we someohe make this idffernt please

i mena i get that "wrote" on a work

but in for example mb, we have the "artist credit on the release"

I think that's from a previous version with no constraints for relationships.

wich is separate from the "various roles through relationships(on works)"

yes

but it's.. very unclear and frustrating becasue I didn't know hat :)

If you edit the relationship by re searching for "The Sleeper and the Spindle" for example, 'wrote' isn't in the available relationships anymore

yea so i'm not gonig ot do that ;)

:D

but my point still stands. "publicist" is a filed in an edition, but "author" isn't. that is really really weird

field*

Agreed. The relationships between entities, and sometimes the entities themselves, aren't quite clear.

anyway what this means in order to link a *book* with a *writer of said book* you have ot create 3, three. thngis: publicatio for the edition, and thne a woork to link edition/book to the autor. not.. ideal

Yes, I found it quite confusing when I first played with the website.

ideally I should only *have* to create the "edition", the publication(editiongroup) to be autocreated or be abl to l ink to an old one :)

i still don't really groc the differnce between a "work" and a" publication" in the grant sense of things. books are often considered the work in and of themselves

I Think There's A Ticket For That™

much like contemporary albums

(where i onte, you have works for the individual recordings, but not a work for the release/album as a whole (unless we're talking concept album or classicla work, but these are not the "album" as a work, but the album is the mendioum of the work yadda yadda))

CatQuest: a short story collection has a lot of works, but isn't necessarily a work in itself - yet it's a publication / edition

Even most classic novels reissued nowadays have at least two works, arguably: the actual novel, and some extra introduction text by an expert telling you about it :)

aha