Open mbjones opened 9 years ago
This project is primarily to support serialization of the package to a local file system such that it may be reliably used by other tools / services. Much of the implementation (e.g. classes for System Metadata, Resource Maps, and Data Object) has been available in the DataONE python libraries for some time - it would be interesting to evaluate the R implementation in that context.
Yeah, in R we decided to make dataone
depend on datapackage
(rather than the inverse), so that other non-dataone R code could also use and create data packages for other systems. This meant that SystemMetadata
was defined in datapackage
, and the dataone
library then imports that and adds all of the service calls. So R's datapackage
is a standalone data package representation that can serialize to BagIt with SystemMetadata and does not depend on having access to any other DataONE services. It would be nice if the BagIt-based package format was standalone on all platforms, even though it means some refactoring. Open to discussion...
Design this python implementation to be compatible with the corresponding R datapackage package. The
datapackage
implementation in R is under development, and has an API that includes classes forDataObject
,DataPackage
,ResourceMap
, andSystemMetadata
. It would be nice if there was a correspondence in the APIs.