Page tree

Versions Compared


  • This line was added.
  • This line was removed.
  • Formatting was changed.


  1. In Review


    serverDuraSpace JIRA

  2. Please squash a bug!


    serverDuraSpace JIRA

  3. Tickets resolved this week:


    serverDuraSpace JIRA

  4. Tickets created this week:


    serverDuraSpace JIRA


  1. Announcements
  2. Sprint checking
    1. Round robin
      1. Ben Pennell
        1. Binary GET, POST done
        2. Binary PUT is in Review
        3. PATCH binary and Fixity checks might be the next steps
      2. Jon
        1. confirmed that FCREPO-3169 is fixed, currently have a PR in for FCREPO-3177 and will be working on FCREPO-3168, FCREPO-3026 and FCREPO-3122 and see how far I get by the end tomorrow
      3. Andrew
        1. 2020 Sprints - Fedora 6#2020SprintsFedora6-Migration-utils
        2. Migration utils this sprint is bug fixes and minor new features
        3. Doesn't know what to do with the .fcrepo directory stuff at the moment in the migration utils
        4. Q: is fedora 6 meant to be able to handle a ocfl repo that wasn't made for fcrepo6?
          1. Yes
        5. Can we start from OCFL copied from one fcrepo to another?
        6. What does rebuildability look like? On demand? Flag that has to be enabled? Command/User interaction?
      4. Peter Winckles
        1. Mostly local stuff
        2. Looked at bugs for resources not being closed in migration util
      5. Danny
        1. Done some refactoring in 3188
        2. PR for getting Archival groups working
    2. Rebuild on start would be a valuable next step for opening the door to revealing what happens next, over CRUD completeness. Danny may focus on this next.
    3. Check in with Peter E about containment and if support is needed.
  3. -
  4. Pairtrees:  
    1. Andrew Woods :  let's avoid creating  
      1. PUT to a/b/c  (where a does not exist :  should it 
        1. create a, b, and c as basic containers?
          1. would this be backwards compatible?
            1. No because it would change the LDP structure of the resources
        2. fail because a and b don't exist?
          1. this would be ideal 
        3. Succeed but create pairtrees
          1. should be avoided if we can
        4. Succeed, creating new resource at created URL, but not filling in missing parents as resources
          1. this would be the best scenario, makes the URIs semantics less important to functionality
          2. introduces complications:
            1. A number of parts of the code base make assumptions based on URI hierarchy.
            2. Situations where client creates a/b/c, and then creates a/b. Would fedora have to change containment of c from a to b? Or fail?
      2. Jared is willing to look at fcrepo4/5 to document existing behavior
      3. Danny concerned with user errors if semantics of uris don't matter anymore
      4. Andrew advocating for determining what degree of backwards compatibility is required, not just doing things because of what previous versions did.
        1. Danny will put together email to gather community feedback, might be helpful to create a document that users can comment on
  5. Build Fedora 6 docker container
  6. Thomas - Could we build a docker at the end of the CI?
  7. Existing docker work: Deployment Tooling
  8. Thomas interested in just a fedora 6 docker container, without the additional environment
  9. Since we are building with Travis, all we need is one or two steps in the pipeline to produce this
  10. Would push the fedora 6 docker container to dockerhub. Might need an organization for it
  11. Thomas will create a few tickets:
    1) setup dockerhub org:
    serverDuraSpace JIRA
    2) update travis with simple docker file:
    JiraserverDuraSpace JIRAserverIdc815ca92-fd23-34c2-8fe3-956808caf8c5keyFCREPO-3193Checking: 14 issues completed
    1. PR remaining for the containment index from Peter Eichman .  Needs some Spring work.
    2. Questions were raised about the containment index and versioning, and versioned containment info will be contained in side-car files.Rebuild PR needs more testing.  Bethany Seeger volunteered
    3. Migration-utils are being tested
      1. There are issues building it in windows that Aaron Birkland has encountered
  12. Next Fedora Demo
    1. Feedback from leadership includes a desire for more realistic fixtures (e.g. more versions contained in OCFL)
    2. Next demo could be persisting binaries and metadata with multiple versions.
  13. Backwards compatibility docs
    1. No ideal solution to handling the removal of pair trees.  Either you collapse the pair trees to keep the containment hierarchy but your uris change, or you convert your pair trees to containers and keep your uris but the containment hierarchy changes.
    2. Another option is to not mint the pair trees at all, and in A/B/C/D, if A contains D, then A/B and A/B/C don't exist.
  14. Docker container
    1. There are issues with the name for the container on Dockerhub.  Thomas Bernhart suggested putting it up under an interim name for the time being.


  •  Clarify in in documentation that multiple simultaneous writes to OCFL are not supported