2018.06.22 Community Meeting

Call-in Details

• 11am ET
• https://duraspace.zoom.us/j/8128353771
• or Telephone:
  • US: +1 646 876 9923 or +1 669 900 6833
  • Meeting ID: 935 807 4182
  • International numbers available: https://zoom.us/u/cex8G1kjQ

Attendees

• Andrew Hankinson (Oxford)
• Rosy Metz
• Sibyl Schaefer
• Peter VG
• Steve Liu
• Neil Jefferies
• Andrew Woods

Regrets

• Simeon Warner (Cornell)

Agenda

1. Brief update from editorial group
2. Community updates
   • UCSD Mellon Grant Proposal
3. Foundational definitions for comment and discussion
   • Definition of OCFL intent
   • Definition of an OCFL digital object
4. Unresolved issues
   • Requirements for inventory.jsonld - (issue-26)
     • Example WIP 1
     • Example WIP 2
   • Decouple storage from OCFL Object - (issue-22)
   • Allow for user-determined checksum algorithms - (issue-21)
     • Suggested points of consensus - plus "it may be helpful to allow OCFL to use other formats for fixity verification"

Notes

Audio recording

• 2018-06-22.mp3

Andrew Woods opened meeting with review of meetings -- Community meeting (once month) & Editorial meeting (biweekly).

Recap of Editorial Group.

Rosy gave an overview of the editorial committee meeting.

• Discussed Open Issues
• Open Repositories recap
• Integration with other repository software

Community Updates

Sibyl presented an opportunity for integration with OCFL and their upcoming Mellon Grant. Need versioning as part of the integration with Fedora, Samvera, and Chronopolis. Rosy and Andrew Woods are part of the specifications team.

Part of the grant is to figure out the technical architecture for Distributed Digital Preservation ecosystem.

Peter would like to see OCFL as a way of easing integration between different platforms, especially versioning of digital objects.

Peter brought up the DAT protocol, similar to IPFS. Designed for research datasets. Will probably be quite specific.

Review of the OCFL object and intent

Sibyl would prefer a stronger definition of OCFL object structure. She agreed to put together a minimal pull request to help wordsmith and clarify.

Peter asked why we didn't use the word 'file' in the object specification, rather than 'content bitstreams'. He agreed to file a pull request to help wordsmith and clarify.

Started with a discussion on decoupling storage from the OCFL object root itself.

Rosy presented a case for storing administrative metadata locally but storing content files remotely. Neil presented a case that an OCFL tree containing only a single version of an object.

Peter brought up content addressable storage as a place to look for a possible solution.