...

1. Follow-up meeting scheduled for next Wednesday
2. Setting context for vanilla-rebuild functionality
   1. Supporting both plain OCFL and Fedora-specific OCFL for read/write, rebuild, etc. would be challenging
   2. If Fedora leaves Fedora-specific data in OCFL directories this might be unexpected/unwanted behaviour
      1. However, this extra data could be stored in a .fcrepo directory which would be fairly easy to remove if desired
      2. The .fcrepo directory may actually be useful for providing context of the previous state of the resources within a Fedora repository
      3. Reaction from OCFL editors: why now use the extensions directory?
         1. Put .fcrepo inside extensions outside the context of the OCFL Object
         2. Risks complexity, splitting metadata in different locations
   3. Fedora should be able to import and read vanilla OCFL without adding specific Fedora data
      1. Fedora would add Fedora-specific data in order to manage/write to the content on disk
      2. Documentation should be clear about what operations on vanilla OCFL objects do and do not add Fedora-specific data
3. Fedora makes assumptions in the absence of specified details based on the platform needs. It seems like there are multiple categories of info that are relevant to different purposes, such as:
   1. common to ocfl objects (timestamps, a checksum, etc),
   2. useful for a linked data platform (interaction models, containment),
   3. fedora specific (external binary location, binary details, creator/modifier details),
   4. user provided RDF/binary content.
4. In the last design document, all of those details were going into .fcrepo except for the last one. Some can be derivable, some only matter in some contexts
5. Proposal: Fedora doesn't know how to upgrade from read-only vanilla OCFL. Upgrading to read/write would require a migration utility
   1. In this case the repository would be in read-only mode - can't create new resources without running the migration utility
   2. Migration would be in-place - adding new versions with Fedora-data
6. Two different concerns: software maintainability vs. long-term preservation
7. Ben C: Desire for basic read/write HTTP layer over OCFL
   1. Doesn't want .fcrepo directory

Fedora Docker Implementation

1. Updated to Fedora 5.1.0, fixed some issues
2. When/how are Fedora snapshots pushed to the Maven repository?
   1. Every GitHub commit triggers Jenkins. Successful builds result in Sonatype snapshots
      1. This is currently broken but should be fixed soon
   2. It would make sense to trigger the Docker pipeline as part of the Travis workflow once Travis has been configured to publish snapshots

Actions

•  Danny Bernstein to ensure that a JIRA exists for scenario of transaction does not succeed completely, but rollback does succeed (see: 2020-04-23 - Fedora Tech Meeting)
•  Ben Pennell to create ticket to use OCFL optimistic locking in transactions (see: 2020-04-23 - Fedora Tech Meeting)

  Jira
  server DuraSpace JIRA serverId c815ca92-fd23-34c2-8fe3-956808caf8c5 key FCREPO-3271

•  Peter Winckles to create a JIRA to investigate support for optimistic locking and mutable head (see: 2020-04-23 - Fedora Tech Meeting)
•  Who:  Clarify in documentation that multiple simultaneous writes to OCFL are not supported
•   Who:  After team has a chance to comment, send Ghost Node idea to general community for feedback
•   Who: Decide who works on what in the next sprint
  

...