Hydra Tech Call 2016-07-27

Owned by Bill Branan
Jun 16, 2017
2 min read

Time: 9:00am PDT / Noon EDT

Call-In Info: 1-641-715-3660, access code 651025

Moderator: Steven Ng

Notetaker: Esmé Cowles

Attendees

Agenda

  1. Roll call by timezone per following order (moderator)
    1. folks outside North and South America
    2. Eastern timezone
    3. Central timezone
    4. Mountain timezone
    5. Pacific timezone
    6. folks who were missed or who dialed in during roll call
  2. Call for additional agenda items (moderator)
  3. Upcoming developer meeting at Penn State (Adam)
    1. Just wanted to let everyone to know that there is a Hydra developer meeting at Penn State in September
    2. Sign up here - https://www.eventbrite.com/e/penn-state-developer-concentrate-tickets-26610150677
    3. Trying to keep it to small groups on three topics:
      1. Bag Import/Export
      2. Generalizing Plum workflows
      3. Dashboard
  4. Slowness with large collections? (Mark)
    1. Mark: Slowness with large collections, adding items to a collection slows linearly with collection size
      1. Minutes to add an item with CurationConcerns 1.1
    2. Trey: Unreleased version (AF master branch) fixes a lot of this, but there is still underlying Fedora slowness
      1. We can't just update CC because there are other issues: ordering of values is no longer stable with updated ActiveTriples/RDF dependencies
      2. Can't backport the performance updates
    3. Justin: So DCE should use AF master to see how much it improves things
    4. Mark: We'll do that and see how much it helps
    5. Justin: May also want to look at AF #1110 — that turns off the indexing of objects as JSON and putting them in Solr
      1. Adam: Should we replace that functionality?
      2. Justin: We can remove it entirely, and use Blacklight find method instead
    6. Mark: Assuming 10K objects in a collection is just too many, what do we do?  Just put a collection identifier in the object?
      1. Esme: We have considered reversing the relationship, so each object would have an LDP container to link from the object to the collections its a member of
      2. Mark: What about user collections?  Maybe one object is in everybody's user collection?
      3. Esme: Maybe we could model those the old way?  I would expect user collections to be smaller
      4. Justin: Batching collection adds can also help
      5. Mark: Yes, we've done that, but it's still taking forever when we finally do add them at the end
  5. Moderator/notetaker for next time:
    1. Moderator: Adam
    2. Notetaker: Justin