#musicbrainz

Hi everyone, my name is harshul jain. I am a pythonist based in India. I will like to contribute to accousticbrainz project using python, flask and postgress. According to https://wiki.musicbrainz.org/Development/Summer... page, I am in process of setting up my development server and understanding how infrastructure works.

So Am I going right?

harshul1610_, yes, that's a good way to start. (also, you were right to ask in #metabrainz, which is the dev channel)

CallerNo6: I don't think it means that a release label is not an imprint, but that we feel an obligation to fill in something for a release label because it rarely makes sense for a release to come out without being associated to a label.

Especially for a major lable.

*label

Obviously self-published stuff is an exception to that.

hawke, yeah, that was going to be my punchline. "A 'release label' is our desire to leave no field empty".

But I stand by my assertion, it is at this point essentially meaningless.

(If 'release label' is "whatever the retailer thinks goes in their 'label' field", then we should just say that.

)

CallerNo6: eh...Does iTunes even have a label field though?

CallerNo6: I thought it was more their 'copyright' field?

Not sure it's even that: "℗ 2015 Mercury Records, a Division of UMG Recordings, Inc." / "℗ 2015 XL Recordings Ltd., under exclusive license to Columbia Records, a Division of Sony Music Entertainment" / "℗ 2015 Sony Music Entertainment" / "℗ 2015 Def Jam Recordings, a division of UMG Recordings, Inc."

CallerNo6: Also I think it's fairly common to guess the imprint based on secondary sources such as copyright holder, catalog number, etc. even on physical releases.

Having a field called 'release label' is a legacy concept that doesn't fit the actual releases we're seeing.

CallerNo6: only if you drop >100 years of releases.

and it's still applicable to physical releases, and many digital releases as well.

100 legacy years :-)

Seems silly to discount those as legacy.

Might as well drop discID support as well.

I disagree. If it applies to releases today, then why is this a contentious subject?

And the vinyl format

and so forth

CallerNo6: It's only contentious for releases that don't actually specify a label.

that's meaningless. what does "specify a label" mean?

er...I'd say have their logo on the packaging, mostly. But you know I consider logo=imprint=label.

Yes, you do. But that formulation only works when it works.

eh, that's probably 99% of the time.

maybe more.

I think it's a safe bet, lol

given that whole >100 years of history thing.

Anyway, I would have no problem with 'iTunes releases are officially considered to have no label' or something like that, but that doesn't make the label field meaningless for all non-iTunes stuff.

No, it doesn'. What makes the 'release label' field meaningless is that:

(Do you really think there's any chance that iTunes has has more than 1% of all releases in history?)

1. we don't use it consistently (because there's no consensus)

2. it's an answer with no corresponding question

(no single corresponding question, that is)

2. "What is the brand associated with this release?" ;-)

associated by whom? itunes? amazon?

associated by the creator of the release.

...that particular question *does* validate the use of the (P) info too, for iTunes and otherwise.

mostly since 'brand' is nicely broad in meaning.

sure, in the sense that "associate" is a loose term.

(and brand :-) )

I think it would be difficult to find a tighter one that covers everything we want it to.

Anyway, I still think it's ridiculous to say "oh, this concept has existed forever, and in the last ~5 years one retailer has stopped using it, guess it's all legacy junk now"

That's my point. Why are we trying to force-fit a meaning to 'release label'?

If it were one retailer, I'd agree. You know that's not what I'm saying (I hope).

legacy isn't junk. Legacy is "we're stuck with an old mental model for a changing world".

because 1. digital releases mean that there might not be a primary source ( no cover for the imprint to be on)

(Though PDF cover art is common enough)

2. releases will continue to become a fuzzy subject, since they can be released incrementally

I definitely agree with both of those, but I think it would be better to adopt/adjust the model only when appropriate rather than dropping the 'legacy' field entirely.

I mean, I suppose you could have a checkbox 'release has no label' or something, but it seems unnecessary.

"drop"? I guess I am kind of arguing for dropping release label, in the same sense that we "dropped" work artists.

if there's data to store, we can probably come up with ARs that capture it better.

As long as assigning those ARs works is basically done the same way as we assign release labels I'm OK with that. But I think that puts us back on square one.

I'm sure I've mentioned this before. A "imprints/logos visible on this release" AR would answer a clear question.

As would "part of <label's> catalog".

Yep.

The second one kinda confuses the issue again though, because a logo (imprint) doesn't have a catalog

well, however you determine the holder of the "catalog".

And the first one potentially because of subtle graphical differences.

But in general, yes.

To summarize, "release label" was once an unabiguous question. As it becomes more ambiguous, we should ask better questions."

Sounds good; how do we get there from here? ;-)

sidenote: sorry, this is what happens when I have coffee instead of tea.

Honestly? I'd like to ask data users what /they/ think the questions are.

(to start)

We have data users?

We don't? Let's get some!

Okay, approaching it from the other side, I guess we'd want to address your concerns, like how similar do logos need to be?

I refer to the "His Master's Voice" logo set as an example. Also "Epic" http://logos.wikia.com/wiki/Epic_Records

I use the data, in some sense. The question for me is "How do I tell this release apart from other releases?"

you guys :D <3

Leftmost, granted, year/label is *usually* a good shorthand for "what edition is this?"

CallerNo6: not as good as you might hope; see the annotations on the undated releases at http://musicbrainz.org/release-group/41757b85-a...

nice

CallerNo6: the annotation also contains a summary of the possible years of release on those

But I'm quite sure there are a ton of releases which have wrong dates due to being unidentifiable repressings.

(or rather, not-easily-identifiable)

Eh, it's certainly a better guess than nothing.

...Which makes me wish we had a better system for data quality.

Nothing new there though

(wikidata's method is kinda close, but perhaps not exactly what I'm thinking of)

Certainly in terms of interface, it is.

what's wikidata's approach? (I really need to learn their system)

(and play with their API and so on)

(under "values" and "rank")

oh, cool, thanks

UI image here: https://www.wikidata.org/wiki/Wikidata:Glossary...

?

Not for any reason. I just like that word.

But now "snak" is my new favorite thing.