As discussed in previous DSpace 7 meetings, our DSpace 7 Preview release is aimed for end of March, followed by a Beta in May (before OR2019). The top priorities for the Preview Release are (in this order):
Consensus that there's only two viable approaches... Either we delete the EPerson entirely (see Pascal's PR), or we use a "ghost user" approach (a special user account which is linked to Items where submitter/Eperson was deleted)
Lieven notes concerns about not catching all the possible NullPointerExpections if we delete EPerson entirely. We have some tests, but we aren't sure we have all scenarios tested.
Pascal notes that it would be a community/team effort to find & fix any that are not already fixed
Tim notes our new test driven development processes should help us to find these issues more quickly. Community Testathon can clean up the rest.
Andrea notes if we go with delete Eperson approach, we should add integration tests to REST API and Angular to check for possible problems (especially around null submitter)
Tim volunteers to help with Integration Tests on REST API
We'd likely need more help on Angular side integration tests
Tim is concerned with "ghost user" approach, as it'd be a specialized EPerson account that would need to be treated differently than all other EPersons. Unclear how much effort that would be, and whether that causes additional complications.
Not a clear consensus on whether we want to delete EPerson or use "ghost user" – different individuals favor different approaches. However, there's no strong objections to either approach from anyone – just personal concerns expressed.
FINAL DECISION: Since we have a PR#2229 and no one has strong objections, let's move it forward as-is. Let's work together to create the missing integration tests (Tim will help) and add comments noting any missing use cases / scenarios where a Null EPerson could be a problem. That way we can build out integration tests.
Skipped discussion of Concurrency in DSpace 7. Will pick that up next week