Issue 11311
Ensure for hosted datasets that relations for "owns" and "serves" are consistently maintained in the registry
11311
Reporter: ahahn
Type: Task
Summary: Ensure for hosted datasets that relations for "owns" and "serves" are consistently maintained in the registry
Priority: Major
Resolution: Invalid
Status: Closed
Created: 2012-06-08 15:44:25.872
Updated: 2013-12-16 17:50:54.376
Resolved: 2013-12-10 17:03:32.11
Description: Background:
Some organisations have a technical installation to host data for other organisations (examples: CanadenSys; GBIF Spain). In these cases, the hosting organisation A "has technical installation" X and "serves" dataset y, while organisation B has no technical installation, and "owns" dataset y. In the dataset aware registry, automatic synchronisation with the technical installation derives dataset information automatically and creates or updates dataset entries.
On the first contact to a technical installation, the synchroniser by default sets both the organisation that "owns" and the one that "serves" the dataset to the same organisation agent, i.e. the one having the technical installation (exception: IPT, where this information is pushed rather than pulled). However, as one of the requirements driving the development of the dataset-aware registry, it needs to be possible to have the dataset linked to an "owning" organisation that is different from the "serving" one. In consequence, this means that
- we need to be able to switch the "owns" relationship of a dataset from one organisation to another, while the "serves" one stays the same
- the synchroniser, in its next run, should not re-generate the switched relationship, either by reverting the change, or by creating a third relationship
- it should still be possible for the registry synchroniser to update the dataset's metadata
- from the information available to be synchronised into the HIT, the HIT needs to be able to index the dataset (under the owning organisation)
Additional question:
How far could the synchroniser be used to point at cases where hosting and owning organisation are not identical? In many protocol responses, this could be deduced from the metadata.]]>
Author: fmendez@gbif.org
Comment: The synchronizer creates relationships only when a dataset is created during a metadata synchronization; if the dataset already existed or was previously created, the synchronizer will never update the relations of that dataset
Created: 2012-06-08 16:01:49.148
Updated: 2012-06-08 16:01:49.148
Author: trobertson@gbif.org
Created: 2012-06-08 16:05:45.652
Updated: 2012-06-08 16:05:45.652
Additional question:
How far could the synchroniser be used to point at cases where hosting and owning organisation are not identical? In many protocol responses, this could be deduced from the metadata.
> this is quite tricky except where direct matches by name are observed.