#metabrainz

I'm trying to make sense of the bookbrainz work/publication/edition hierarchy. Is there any guidance yet on what counts as the same publication versus different publications?

related: what's the best way to handle introductions? I was thinking I'd make a work entry for the introduction, and link it in the publication with 'contains', but there doesn't seem to be a relevant type (nonfiction maybe?) and it would mean the common case of later editions with a new introductions are different publications

which is maybe okay?

tungol, are you familiar with the frbr model?

or with the musicbrainz model?

bookbrainz is similar

I've edited on MB for a while, yeah. Not sure what frbr is.

from a discussion the other day, "publication" may become "edition group" if that helps :-)

(I took that to mean that a "publication" is analogous to an MB-work-group

)

so by analogy with MB, an edition of two distinct works together would be a new publication, but not an addition that adds a new introduction, which is more like bonus tracks

I think that sounds right.

I think these definitions are still being developed (from what I could tell the other day)

At this point, Leftmost will step in and explain what I got wrong :-)

oops, "mb work group" was supposed to be "mb release group". I have no idea what a work group is. grr.

yeah, I figured that's what you meant

is edits just hanging forever when you try to submit them a known issue? That keeps happening to me, semi-intermittently

tungol, there's nothing formalized as far as what constitutes a new publication. Right now we're hoping to drudge up some edge cases to see where we need to put in guidelines. Your MB analogy is apt.

It's kind of a known issue. It means that something went wrong in making the edit and our error reporting is terrible. Is there any pattern you've noticed in when it does it?

no pattern yet, but when I do hit it just retrying the edit in a different browser doesn't fix it. Last time I created an author and it didn't save the name, and trying to go back and add the name would just hang. This time I'm trying to add an edition to a publication group and it doesn't want to go

ah, okay yep

I was

so in general what level should works be linked in with the "contains" relationship, publication or edition? Publication makes sense for the main work, but if I make a work entity for the introduction that would need to be at the edition level.

and having different works linked at different levels feels a bit weird

Edition.

Relationships are going to be revamped soon to restrict entity types and "contains" will be Work-Edition.

Err, Edition-Work.

good to know

are there any edition-edition relationships? the book I'm using as a test case right now is a full reprint, including title page and publishing information, of an earlier edition, with their own title page and publishing information, plus a introduction, added beforehand, it uses the phrase "complete photographic reprint"

I'm thinking there's no way to represent that right now

No relationship like that yet, but you should file an issue for it (if you don't mind).

sure, I can do that

I'll make one for a work type appropriate for use with introductions as well, if you think that makes sense

i went through scraper.js and i now know how it works. (my first guess was close enough to the actual implementation)

i went through scraper.js and i now know how it works. (my first guess was close enough to the actual implementation)

and i have an additional idea.

since the users wont have much technical know-how we can ask the user to select the div consisting of the songs listed.

alastairp: you said we cannot ask user for their lastfm credentials, then how r we going to fetch records using api? we sure need credentials/api-key

alastairp: my idea is to scrap the key by parsing DOM and then make api-calls

armalcolite: if a user's scrobble data is public, then we don't need *their* API key. We only need *a* API key

We can embed it in the script file that we serve to the user from the bookmarklet

alastairp: ok. what about private one?

alastairp: how about we build a add-on similiar to last.fm's and let user access their data?

what do you mean an addon?

like a chrome extension?

Freso: there's no "social area" as such in Discourse, is there? As a category, I mean

(I wasn't sure whether my last post should have a category or not :p)

alastairp: took a look at bigtable

the schema design principles seems like redis on steroids

"These tables will grow at the rate of approximately 2 billion rows per day, which Cloud Bigtable can handle without difficulty."

uh?

redis is a memory store.. bigtable-like things are... not?

yes but it's a key/value store

bigtable is much more like cassandra than redis, imo :)

the priniciples of designing your db schema is the same

haven't used cassandra

I was comparing bigtable and influxdb

alastairp: I tried to work out the details for the job evaluation project and have replied to your comment.

where influxdb is "smarter" with timestamps

Please have a look when possible and let me know if I missed something.

while bigtable just uses timestamps in keys

but the sorted indexes make it much more appropriate for acousticbrainz

if it can handle ~8000 rows for every song

which it claims it can

this amounts to approximately 28 billion rows

and it took us a year to get to that, so we're not growing at 2b/day :)