MBS-12977: Correctly specify the licenses of the JSON dumps #2897
Conversation
reosarevok
force-pushed
the
MBS-12977
branch
3 times, most recently
from
245a9f1
to
cec5839
Compare
We were claiming this is all CC0, but that is not correct. The JSON dumps contain annotation, rating and tag data, all of which is supplementary / CC-BY-NC-SA data. Talked with mayhem and mwiencek about this, we agreed that the right thing to do is to update the dump docs to specify the right licenses rather than dropping the supplementary data, since it can be useful and there's no great way to provide it separately like on the MB dumps.
|
Hi! I really think this is not a good approach at all -- unless we wish to be a bit evil about it, in which case its fine. 😈 The problem I see is that now in order to consume this dump, which is intended to be ingested into a document store, the user is required to actually process some of the dump files and write a filter to remove the non CC0 bits before they can import the data. Far less than ideal. But if we want people to not notice this (which will be easy) and allow people to import the non CC0 zero data so that we can sue them later for doing so, then this is a great approach. While we really don't need more data dump files, it would be best to put out a sanitized version of the files that contain non CC0 data and label them as such, so that when people download the files they have to make the correct decisions right on the spot. |
|
I mean, enforcing this would be complicated anyway given we have given the whole data as CC0 for years, so if someone stopped importing the supplementary data tomorrow and kept all the existing ones until this change, they would be in the right. Having two dumps might be doable, as long as we can do it without it doubling the time it needs to generate the dump (since that's already very problematic as it is, taking over a day). |
|
Converting to draft since we should look into dumping two files, @mwiencek |
Fix MBS-12977
Problem
We were claiming our JSON dumps are under a CC0 license, but that is not correct. The JSON dumps contain annotation, rating and tag data, all of which is supplementary / CC-BY-NC-SA data.
Solution
Talked with @mayhem and @mwiencek about this, we agreed that the right thing to do is to update the dump docs to specify the right licenses rather than dropping the supplementary data, since it can be useful and there's no great way to provide it separately like on the MB dumps.
As such, this moves from having one COPYING file to two, one per license, with the README file telling people what content is covered under which license (I also updated the docs with the proper licensing info.
Testing
Added tests to make sure the license files are in the dumps and the README tells people to check them.