Page History

...

Notes

Josh Westgard - UMD Update

• Obstacles for accessibility?
• Currently working with reduced staff so not able to address it right now
• Sanitized html is possibly more accessible than image files as a bonus
• What’s your solution for displaying IIIF resources?
• we currently run a Loris server, and use a modified Mirador for display in browser
• We have a custom rails app (pcdm-manifests) that turns fedora objects into IIIf manifests
• https://github.com/umd-lib/pcdm-manifests
• https://github.com/umd-lib/mirador

Quinn Hart - UC Davis Digital Collections

• PCDM structures - moving away from these?
• Usedschema.org to help organize data
• Service extensions? How are they implemented (API-X?)
• Inspired by API-X, doesn’t use it but built similarly
• Fedora has services directory, so use this to add additional services you want to see
• Interaction w/ fedora through LDP

Scott Prater - Migrating Fedora 3 to Fedora 6 Using Migration-Utils

• https://youtu.be/Z-A2Ob2mxPI - Migration video
• Does the validation tool verify checksums?
• Yes - have talked about making it optional as it can impact speed
• Released on main branch currently and there will be an upcoming Alpha
• In your repo with 10M objects are most datastreams / binaries internal or external?

NLM Repository Update

• Migrated object - actual image files are not included in this file?
• Using external data streams from Fedora
• Fedora 6 - how permissions will work with WebAC? Do all pieces need to be in one container?
• Still exploring it and only providing a reference in Fedora 6
• Don’t have exact same use case for permissions in their system so organize binaries externally
• So, I understand you have on the order of 5 million objects, but the public Blacklight UI shows about 145k.  Are you using a filter for the indexing?
• 5mil pages + still images, videos, born digital items - each page is an independent object
• Do you have to do much work with an LDPath transform to set you the Solr index?
• Not much to do with LDPath

• Not concerned about turning off checksum validation for migration efficiency, but it is critical to validate 100% post-migration.  You might check with the digital archivist community, but I understand it to be a cornerstone of responsibly accounting for provenance of every file for librarians and archivists.
• You can turn it off to migrate, then run migration validation afterwards

• UNC we didn’t end up using webac since our materials are not hierarchical from the perspective of fedora. Webac would have needed to follow membership relationships, but I can see how enabling that could be problematic given the possibility of multiple membership

Notes from Chat:

James Creel to Everyone (4:19 PM)

Do you have to do much work with an LDPath transform to set you the Solr index?

Joshua Westgard to Everyone (4:20 PM)

Thanks, Ben. This has been our experience with permissions.  We can't rely on PCDM relationships because of multiple membership and we didn't organize our content hierarchically according  to LDP containment, and as a result webac is a challenge.

Josh Westgard to Everyone

We ended up implementing our own access control approach outside of fedora, which has worked fairly well for us, but is obviously not very shareable. But I don’t think it’s unreasonable to propose membership based webac inheritance, it would likely require configuration to define which properties should be followed, but it may be possible.

Day 2

Samvera Update

• What other backends are being explored in Valkyrie?
• Postgres? The idea is that any backend could be moved in once the Hyrax Valkryization is complete
• UC Santa Barbara would be somewhere to look at for more details
• Fedora 6 testing - Hyrax in current release doesn't yet support through Valkyrie?
• A little more work to be done there yet with the Hyrax/Valkyrie interactions
• Berlin State Library is using a Samvera front end and have been doing some migrating from Fedora 4/5-6
• Hadn’t we validated with Fedora 5 that it worked with Hyrax?
• Originally thought they would need to build an adapter to move from 4-6 but looks like they won’t
• Hyrax 3 just released - what were the newest updates?
• Support for Rails 5.2 and Ruby 2.7 are added; we now recommend Rails 5.2 for all applications.
• We dropped testing for Ruby 2.4; we now recommend at least Ruby 2.5 for all applications.
• Out-of-the-box metadata schemas are updated to more closely align with the recommendations of the Samvera Metadata Interest Group.
• File versions and transparent PNGs are now supported by our IIIF installation. Universal Viewer's install process is improved to use yarn.
• Valkyrie is introduced as a production dependency and a custom adapter (Wings) is introduced to handle backward-compatible use of Valkyrie within Hyrax.

Islandora Update

• Have you considered rolling out Fedora in containers and using static containers for load balancing
• Almost there - ISLE uses Docker compose so it just kind of works, but certainly very interested in it
• Kubernetes roll out - of big interest to users
• In terms of migrating - are you working with the migration paths project?
• Whitman College is an Islandora school using I7, Drupal 7 and Fedora 3
• Need several layers of metadata remediation in the process
• Converting things from MODS to CSVs
• Once done, will run Islandora workbench to move data

James Creel - Texas A&M University Libraries Fedora Updates for 2021

• Struggling with content models based on RDF stored in Fedora
• How do you manage those different content models that you’re applying?
• Happens in different steps
• Ingest using in-house tool MAGPIE
• Configure indexing in Fedora with LD path transforms
• IIIF then available and known schema available

Fedora Migration Break-Out

• Will be releasing new and improved Camel toolbox with Fedora 6
• Community wanted something so as not to be tied in to Karaf
• NLM wanted to see how data coming out of Solr would look in Fedora
• Should they try it in Marmotta?
• Sideloading - is it in 6?
• NLM this seems like an attractive use case
• It is available and works
• Does perform validations on OCFL objects to ensure it is recognizable
• Not practically usable at the moment - probably better done with some other type of utility
• The feature as it exists only supports adding new objects, not for old objects
• Archive moves and OCFL
• When doing migration from 3-6 b/c of the way OCFL structures archival groups, but maybe you can explain it further
• New feature of 6 - you have some control over where files live on disk
• Community wanted some way to co-locate all those objects
• When you use archive groups, everything gets versioned together as one unit
• Can you have a mix of archival groups and single data stream OCFL objects?
• Yes, because they functional independently
• However when you modify objects in Fedora it locks to that single request

Resources