#musicbrainz

is there some kind of wiki page or similar for tags?

seeing which are popular, agreeing on certain tags for certain uses, etc

seeing game,game music,vgm on the same RG doesn't make me too hopeful :P

top tags are listed at https://musicbrainz.org/tags

documentation is available at https://musicbrainz.org/doc/Folksonomy_Tagging

I was hoping for a community effort to agree on tags for when synonymous tags exist

there is no guideline, it's open, community agreement is expressed with vote

dominikh: if you mean for genres, we have one "marked" genre, while the others currently aren't

But for non-genres, there's nothing specific, no

reosarevok: right, for non-genre. I just noticed a lot of duplicate tags, like the aforementioned game,game music,vgm – which all express the same idea

so of course part of me wishes for that data to be normalized :P

An option would be to make a bot that adds the other two when one is there, then it's "normalized" ;p

bonus points for nefariousness if it also downvotes the other two?

-1 to the nefariousness

but the other idea i propose hartily

I'm torn on that idea. It does mean that if the original tag was wrong, and nobody has downvoted it yet, the bot would end up adding another incorrect tag

True. But tags are not right or wrong, you just agree with them or not :p

If I think Bach is video game music, then it is! :D (for me) - maybe I'm making a video game that uses Bach :p

in that case I propose a plethora of new relationships so that we can record the absolute truth :P

Even Wikidata doesn't try to do that and they have allll the relationships :p

hm, question about the actual database: when entities get merged, redirects are being added to _gid_redirect tables. does that imply the original rows with the old gids get deleted?

Yeah

But relationships etc are transferred to the entity they got merged into first

right. I'm playing around with some software/database designs that rely heavily on MB data, using a local version of the database. if entities can be deleted, I'll have to keep that in mind when designing my foreign keys :)

^ entities can *also* be outright deleted.

does that ever happen to entities that aren't spam?

ideally "hardly ever", but it can happen.

dominikh: it also happens if someone just adds, say, an artist that does exist, but doesn't use it

(adding at least one relationship, even if it's just a link, avoids that)

oh, good point. I forgot about that.

right. those cases are probably the easiest to handle with my use case. if nothing uses it, I won't, either. merges prove far more difficult. imagine an experimental music database/player that pretends to live in a world where everything is tagged with musicbrainz IDs.

files are stored as their track IDs

a bunch of postgresql triggers can probably take care of it, though

Cool. I'd like to see how that comes together.

unlikely to turn into anything user-friendly, on account of all the databases that need setting up and maintaining, but it may be a nice toy for the nerds among us

I guess one other (hopefully rare) case you'd need to consider is when something is merged in error, and then needs to be recreated (with a new MBID).

that pretty much requires manual intervention and retagging, won't it

not always, but yeah

e.g. recordings might be incorrectly merged, but the release ID remain unchanged

yeah, I'll probably only store track IDs and figure out the rest via joins. reduce the risk of IDs disappearing

at any rate, early days. will see how many issues I run into

does anyone see a reason why https://musicbrainz.org/release/1299f319-8ff4-4... and https://musicbrainz.org/release/ac55174d-3965-4... shouldn't be merged?

if not, i'll open a merge edit

thats one old release

https://www.youtube.com/watch?v=s4BJ3wng6Mk wow

ftr: that described merge edit is now open: https://musicbrainz.org/edit/57782481