Dataset licensing metadata about dat

[max-mapper]

agreed. I'm considering building on top of http://www.dataprotocols.org/en/latest/data-packages.html which is inspired by package.json. running dat init should probably go through a wizard like npm init and set a default license (NPM sets bsd by default)

[junosuarez]

A few thoughts: headers and metadata in general are very important for reasons of organization, dataset discovery, licensing and authorization, specifying encoding, support contact information, etc. However, static representation of this, for example the datapackage.json descriptors, works great for snapshots (like semver published versions, a la npm) or even continuously versioned sources, like a git repository - after all, you just get the representation of the descriptor file as of a certain point in time.

For live-replicated data sources, there needs to be a mechanism for handling (and replicating) changes in metadata. This could be as complicated as interleaving append-only patches to metadata in the main replication channel, or another way comes to mind: TTL timeouts, like DNS. People getting a one-time data dump can get the descriptor as of that point, and clients that are continuously mirroring a data source can refresh the descriptor resource as often as necessary.

I realize I'm replying to an older thread - please point me in the right direction if this has been superseded by another discussion thread.

[junosuarez]

Also, rather than specifying or establishing conventions for all possible metadata, including jsonld for semantic metadata would be supercool. A core "minimum required" set is fine, and what the datapackage.json spec includes is reasonable.

[okdistribute]

I think we can close this, we have a package.json now in dat and people can add a license to that (i.e., npm).