- Andrew Woods
- Rosalyn Metz
- Mark Jordan
- Jennifer Gilbert
- Peter Winckles
- Aaron Birkland
- Christopher Awre
- Ben Pennell
- Melissa Anez
- Doron Shalvi
- Ensure alignment between Committer and Leadership teams on technical priorities/direction
- Establish date (or criteria for date selection) for initial F6 Alpha release
- Resolution from previous meeting
- Use cases that will be supported
- The repository administrator will be able to create an OCFL object (with external tooling) in Fedora's OCFL storage hierarchy according to Fedora's expectations of where such an object should exist and how the object is to be structured.
- The repository administrator will subsequently be able to either reindex the entire OCFL storage root or request that Fedora index a specific OCFL object.
- Design - Fedora OCFL Object Structure
- Design - Fedora Header Files
- Use cases that will be supported
- General Fedora 6 updates
- Leadership/Governance updates?
- Primary focus of f6 is swapping OCFL for Modeshape: by and large that is done. Getting from Fedora 3 to Fedora 6 is there as well; there is significant testing going on; NLM migrating approx. 3 million objects.
- Took about 7 days for NLM to migrate. Seemed reasonable to them, but this is the time to discover issues and focus on optimization
- Is their migration generalizable? Good to know where the variations lie, and what impact on migration time
- Currently there is no support for Fedora 4 or 5 migration support to 6. This is vital. We do not want to create another Fedora island; do not want a dead end for folks on fedora 4 and 5.
- No direct sense of timing, but we do know that we're not releasing fcrepo6 without this
- From a leadership perspective, the intent of 6 was "everybody moves to 6 equally"
- How much effort is 4 or 5 to 6?
- We can export right now to "something that makes sense". The idea is to do that, then as a subsequent task migrate that export into OCFL.
- We don't really know scale of fedora 4 and 5 repositories in the wild, don't know if there are scaling issues with existing migration tools
- No export→OCFL tooling exists right now.
- Anybody with fedora 4 or 5 data to migrate?
- JHU: Has fedora4, 300k-500k objects, ACLs, some large containers. Not a good use case for migrating to archival groups
- UNC probably has 1-3 million fcrepo5 objects before ready to migrate to Fedora 6. Interested in migrating to archival groups (as their fcrepo5 model is more or less an emulation of fcrepo3 patterns)
- Leaders make a point of clarity: we're not releasing 6 until all migration tools are ready. Make it clear that the absence 4→6 and 5→6 tools will be a blocker to a release if fcrepo6 is done, but migration tools aren't.
- Don't release fcrepo6 alpha until all migration tool is ready? if 4, 5 migration isn't ready, we haven't met the core objective of 6.
- Andrew: thinking migration tools are a requirement for beta, not alpha. Some value in starting to get feedback with an alpha ASAP, regardless of migration tools?
- Folks have been using 3→6 migration without us knowing.
- Don't think it's vital to decide right now on alpha? Discuss at leaders meeting?
- At the very least, 4→6, 5→6 migration tooling ought to be a higher priority than it has to date
- Need people with machines to run tests! Tests themselves are ready to go
- Is there anything we should be doing more broadly to encourage folks to test?
- Defer to product technology group to lead this thread?
- Start slating 4→6 migration work into sprints? Next sprint is coming up soon.
- Do we have any thoughts of Alpha timing? Thinking "some time this year"?
- How many sprints for migration tooling?
- Not clear how much developer effort is available Team is small but enthusiastic.
- Not prepared to make a time commitment.
- How many hours of development for migration tooling?
- Andrew thinks three sprints of 1-2 people focused on migration.
- Work is well-specified
- Feature tracking page should be refactored a bit to be more easily consumable. Just focus on top bullets for now, look for fields of green, red, blue.
Resolution from previous meeting
- Currently, there is "reindex everything" available, not "index these few novel objects" (yet)
- Recipes for creating fcrepo-compatible objects available
- Side loading OCFL will be the basis of importing to fcrepo6, if not through Fedora's API.
- Expectation is that it is a fedora object, not just vanilla OCFL
- Fedora 4/5 Migration tooling must be included in the Fedora 6 release; indeed it should be included in the Alpha release for Fedora 6
- Fedora 4/5 Migration tooling has been well specified and work on 4/5 migration tooling should become a priority
- Type your task here, using "@" to assign to a user and "//" to select a due date