#musicbrainz

/

      • johtso
        like the lyrics project
      • reosarevok
        Heh
      • Well lyrics are fucking mess
      • johtso
        I can't bare to add relations to recording by embedding links in wiki markup..
      • reosarevok
        (in general)
      • We'd love to do lyrics ourselves for example, but licensing is ridiculous
      • :)
      • So I suspect most lyrics stuff has enough problems remaining alive to care much about best practices and design
      • LordSputnik1: well, bookbrainz to cocktailbrainz is still a big one! :p
      • johtso
        urgh.. licensing.. copyright.. the enemies of data preservation
      • reosarevok wants a chocolatebrainz
      • reosarevok
        There's no decently done chocolate DB out there that I've seen :p
      • johtso
        but is it a series of chocolates?
      • reosarevok
        Well they're probably released
      • LordSputnik1
        reosarevok: Well, that depends on your entities... you'd probably want a Glass, GlassType, Drink, Vendor, and then it's just a case of updating the schema
      • reosarevok
        :)
      • johtso
        it did come in a nice plastic box after all..
      • reosarevok
        Well, arguably the same is true of MusicBrainz, LordSputnik1 ;) But designing separate schemas for stuff is hard :)
      • LordSputnik1
        reosarevok: that depends on your entities, and how you go about storing data - if you move more of it into relationships, and focus on storing only an entity's properties on the entity itself, then changing from one Brainz to another is much easier
      • reosarevok
        Oh, sure, that much is true - wikidata is much more flexible than MB
      • (by basically storing only a name on the entity and everything else as properties)
      • It still requires a lot of schema definition to make sense of the stuff, but not so much coding to make the schema work :)
      • LordSputnik1
        Also, you could argue that by generalizing MB entities and storing them in a sharedBrainz, then you'd have a lot less entities to redefine between Brainzes
      • reosarevok
        I mean, going full wikidata style is interesting in itself
      • They have just one entity "type", which can then be defined as a property, so basically is:artist
      • Our stuff has its benefits, but I'd be curious to see MB running on the WD model and see how it fit
      • Only without the deletionists :p
      • LordSputnik1
        The code to only allow certain relationships for certain types would become more difficult, certainly
      • reosarevok
        Well, to a degree, yes
      • You gain flexibility in some parts by making others more complex
      • LordSputnik1
        We have that with BB already, due to our slightly different model of entities
      • reosarevok
        I'm not saying we *should* work like that, just that it would be interesting :)
      • LordSputnik1
        Ideally, entities would only store "adjectives", according to http://en.wikipedia.org/wiki/Entity%E2%80%93rel...
      • reosarevok
        Of course, even with how we work now, that's still true
      • ("orchestra married choir" is nonsensical, yet perfectly enterable in MB right now)
      • LordSputnik1
        Yeah, that's where it might help to eventually have "Person" and "Group" general types, rather than "Artist"
      • we've talked about this in bookbrainz-devel a little (although sticking with Artist/Creator for now)
      • reosarevok
        I mean, in BB "group" or "collective" is probably less likely to appear
      • (or well, less *common*, it will certainly appear)
      • (but if it's rare enough, trusting your users' common sense is less of a problem :p)
      • CallerNo6
        I protest. The bible says 'Adam and Eve', not 'Adam Mansell Orchestra and the Stevens Street Worship Choir'
      • LordSputnik1
        :D
      • johtso
        it's pretty confusing add relationships that don't give any sense of direction.. like "sibling"
      • reosarevok
        Well, in that case the direction just doesn't matter :)
      • It's just there because there must be one, with the way MB works right now
      • johtso
        well it does if you want to avoid relationship webs
      • or whatever the term was
      • reosarevok
        Oh
      • I think for siblings, most people decided that "shrug"
      • johtso
        I thought the standard was to relate all siblings to oldest sibling
      • reosarevok
        Well, yes, that's the official guideline
      • But I think barely nobody has been following it much
      • johtso
        ah, there's the guidelines.. and then there's reality :)
      • reosarevok
        In any case, even if you do it that way, it doesn't matter whether the oldest is entity0 or entity1 in the DB
      • That just says "do not add a sibling rel between the other ones"
      • johtso
        oh really?
      • that's good to know
      • reosarevok
        I mean, dunno if it might to some purist, but given the lack of following for the loosest interpretation :p
      • johtso
        right, the releationship entity doesn't contain any sense of direction so it doesn't matter who's 0 and who's 1
      • reosarevok
        Yup
      • LordSputnik1
        :P I think it'd be possible to store all siblings in a single relationship in BookBrainz with a few small tweaks
      • reosarevok
        I do have a ticket to kinda give up on sibling and make do-not-cluster be for other things (like remasters)
      • LordSputnik1: the goal is to be able to define rels better in MB too, using roles
      • LordSputnik1
        What are roles?
      • (in this context)
      • reosarevok
        (so "sibling" is "n siblings", "member" is "1 group, n members", "recording" is "1 recording, 1 place, n performers, 1 recording engineer, etc"))
      • (instead of all being A x B)
      • But it's a lot of work :)
      • So for now...
      • gnu_andrew joined the channel
      • Well, dunno if it's an official goal, anyway, but it's been talked about in the past a few times and the main objection was amount of effort but not actual opposition to the idea. At least in the core team
      • But yeah, pie in the sky at the moment I'd expect :) Who knows in the future though
      • LordSputnik1
        Yeah that seems like a lot of work
      • reosarevok: BB relationship editor is sort of working here, have a play if you like/have time - http://bookbrainz.org/creator/09688598-bca7-44c... :)
      • (although if you try to submit, it'll silently fail because you aren't logged in...)
      • reosarevok
        Heh
      • can't right now, I really should be in bed so I'll just finish work
      • Remind me tomorrow or one of these days though :)
      • LordSputnik1
        ahh, BBC stuff?
      • reosarevok
        Yup
      • Almost done for the day, but my gf has been looking at me like "bed time? please?" for a while so I should stop doing random stuff :p
      • LordSputnik1 has left the channel
      • Nyanko-sensei joined the channel
      • ruaok joined the channel
      • pbryan joined the channel
      • JesseW joined the channel
      • Wheeljack joined the channel
      • drsaunde joined the channel
      • JesseW joined the channel
      • skd5aner joined the channel
      • hibiscuskazeneko joined the channel
      • STalKer-X joined the channel
      • JoeLlama joined the channel
      • KRSCuan joined the channel
      • shredpub joined the channel
      • JesseW joined the channel
      • simukis_ joined the channel
      • hopphopp joined the channel
      • hopphopp
        does anyone recognize this error http://pastebin.com/caEYM3gM ?
      • i just got it when I tried to save updates to an album
      • Junior_ joined the channel
      • simukis_ joined the channel
      • simukis_ joined the channel
      • nikki joined the channel
      • chungy joined the channel
      • zas joined the channel
      • ruaok joined the channel
      • reosarevok joined the channel
      • reosarevok joined the channel
      • simukis_ joined the channel
      • hibiscuskazeneko joined the channel
      • shredpub joined the channel
      • DarkerAudit joined the channel
      • DarkestAudit joined the channel
      • DarkerAudit joined the channel
      • DarkestAudit joined the channel
      • sertansenturk joined the channel
      • ruaok joined the channel
      • achadwick joined the channel
      • KRS-Cuan joined the channel
      • Freso
        "johtso | tools that consume musicbrainz data don't understand the concept of a series" - doesn't mean they can't be made to, so not really an excuse. MusicBrainz should not worry about how it's data can or cannot be used in other applications, as long as it is making its data available for use.
      • reosarevok
        Freso: I mostly agree, and I'm a data purist, but I do sometimes feel that if we're restricting it so much that it makes it completely useless and very hard to make usable, it might be too much :) )
      • I mean, in that we need people to have some kind of reason to add stuff, or it won't get added at all
      • So I'm fine with not being too strict on grey areas (although still strict on things that are just wrong)
      • Freso
        johtso: I deliberately make sibling relationship clusters. Since, 1) I don't always know the age of all siblings, 2) If I go to the non-oldest sibling, I can still get information on all siblings, without relying on fetching an "entire" different artist+relationships.
      • johtso
        Freso: yeah.. the current implementation is definitely flawed in that respect
      • Freso done with backlog o/
      • is the optimum solution to have some kind of siblingship entity that the various artists are related to?
      • Freso
        Isn't that basically what reosarevok talked about wrt "roles"?
      • johtso
        ah, skipped over that *reads back through the discussion*
      • reosarevok
        Yeah, that's the optimal choice :) Just not trivial, so for now we do what we can :)
      • johtso
        wow, that sounds pretty snazzy
      • nikki
        I think pretty much everyone links siblings to all the other siblings, it's not the most ideal way to enter the data, but it's not like any of the currently possible options are great
      • johtso
        As long as the data's there it can always be migrated :)
      • nikki
        it's just that every time we tried to fix that damn guideline, someone complained :P
      • so people were like fine, whatever, we'll just continue not following it like we have been since forever *grumble*, or something
      • chirlu` joined the channel
      • chirlu`
        The “siblingship entity” to which all the siblings should be related to is generally known as “parent”.
      • -to
      • nikki
        people don't really like adding placeholder artists though, it feels like a hack
      • (and even if you disagree, that won't make people do it)
      • chirlu`
        And introducing a new siblingship entity is better?
      • reosarevok
        Not a new entity, no :)
      • The idea would be to allow relationships to have roles, so the sibling relationship wouldn't be A sibling B but "siblings: A and B and C and ..."
      • chirlu`
        The “relationship that can have N participants” idea is broken, too.
      • reosarevok
        Why? :)
      • chirlu`
        For several reasons (and I have a tab with MBS-1159 open since a while back to comment on it).
      • reosarevok
        In any case, that only would suggest even more that what people do (link everyone) is the most reasonable choice
      • But I'd be curious to know about the issues, really, because relationship roles seem pretty sensible to me