Dataset modification and versioning
The fundamental principle of the PDL is that datasets are stored unchanged for an indefinite period. Should changes to a dataset be required then this will be implemented by publishing a new version of the dataset, which will involve the following
- The assignment of a new version number — a simple integer sequence that does not support the concepts of major and minor upgrades.
- The assignment of a new Digital Object Identifier (DOI).
- The creation of a landing page for the new version of the dataset that includes its full version history.
- The modification of the landing page of the previous (but not earlier) version of the dataset to provide a link to the new version.
- Storage of the new dataset in addition to previous versions.
The one exception, to cater for data sets that acquire new data whilst the existing data remain unchanged, is the ability to 'grow' a DOI. In this case it is permissible to add data files and documentation to an existing published data set under the existing DOI as long as the original data files and documents remain totally unchanged.