Task #853
Story #927: R Client needs DataPackage model and implementation
add R client DataPackage constructors from data or science metadata
100%
Description
Currently the R client assumes it is getting a data object, and so the constructor for [[DataPackage]] adds the retrieved object to the [[DataObject]] data list. Instead, need to inspect the [[ObjectFormat]] of the incoming object form system metadata, and use this to determine how to construct the [DataPackage]. Also, in some cases we may be able to use lazy loading to grab the new parts of the data package as needed -- ie, only download science metadata or data objects as needed by calls to [[DataPackage]] accessor methods. This might be based on the size of the object (fine to cache small objects).
Under the new design (ORE packaging), the behavior probably changes somewhat - given a pid, it needs to make sure the format of the pid is that of a ResourceMap (RDF/XML), and throw an exception if not so.
Lazy loading leads back to expanding the underlying Java libclient D1Object class to be able to represent the object bytes in a flexible manner - download should hold a "D1RemoteDataSource", for example.
History
#1 Updated by Matthew Jones about 14 years ago
- Target version deleted (
CCI-0.6) - Milestone set to CCI-0.6
- Parent task set to #927
#2 Updated by Dave Vieglais about 14 years ago
- Start date set to 2010-10-07
- Tracker changed from Bug to Task
#3 Updated by Dave Vieglais about 12 years ago
- Target version changed from Sprint-2012.39-Block.5.4 to Sprint-2012.41-Block.6.1
#4 Updated by Chris Jones about 12 years ago
- Milestone changed from CCI-0.6 to None
- Assignee changed from Matthew Jones to Rob Nahf
Assigning to Rob.
#5 Updated by Matthew Jones about 12 years ago
- Target version changed from Sprint-2012.41-Block.6.1 to Sprint-2012.44-Block.6.2
#6 Updated by Rob Nahf about 12 years ago
- % Done changed from 0 to 40
- Description updated (diff)
#7 Updated by Rob Nahf almost 12 years ago
- Target version changed from Sprint-2012.44-Block.6.2 to Sprint-2012.50-Block.6.4
#8 Updated by Rob Nahf almost 12 years ago
For a given pid, getD1Object can return either a data object, metadata object, or resourceMap object (the packaging mechanism)
The desired behavior is probably to return a DataPackage, in order to get the appropriate functionality, but that requires downloading all of the member objects, which could be unexpectedly memory and time intensive.
Implementing this should probably wait for a refactor in d1_libclient_java to lazy-load (proxy) the content of the member objects, which mitigates this issue. Alternatively, an option could be added to "download all", or provide a list of identifiers of which to download, as well as create a downloadMembers method to allow a more interactive approach.
#9 Updated by Matthew Jones over 11 years ago
- Status changed from New to Closed
- % Done changed from 40 to 100
- translation missing: en.field_remaining_hours set to 0.0
Agreed -- this feature should probably wait until the underlying libclient supports lazy loading. So, the support for DataPackages can be considered complete for the purposes of release 1.0.0. Closing.
#10 Updated by Matthew Jones over 11 years ago
- Target version deleted (
DataONE R Client 1.0.0)
#11 Updated by Matthew Jones over 11 years ago
- Target version set to DataONE R Client 1.0.0
