Samvera Tech Call 2017-05-31

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: Steve Van Tuyl

Attendees:

Agenda

  1. Roll call by timezone per following order - ensure notetaker is present (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

    7. Welcome all newcomers!
  2. Agenda (moderator)
    1. Call for other agenda items (moderator)
    2. Deprecate Hyrax's Noid functionality (justin)
  3. Moderator/notetaker for next time (moderator)
    1. Moderator: 
    2. Notetaker: 
  4. After call, this week's notetaker should create the agenda for the next call.

Notes

  • Roll Call: super!
  • Agenda:
    • No new items
    • Deprecate Hyrax Noid Funx (justin coyne)
      • when we do big ingests into hyku, af noid minter is a bottleneck (one connection at a time, please)
      • can we give up noid in order to make ingests quicker?
      • JC made change to AF Noid so that if you want to use it you can include it
      • this helps loosen the coupling between noid and hyrax
      • PR in hyrax now - slated for removal in hyrax 3
      • Discussion:
        • Q: How many people using hyrax want to use noids vs. not?
        • If just a handful of people want to use noids, then go ahead, but if most people want noids, then keeping them around and making it turn-offable (in configuration) might be better
        • making this flippable results in objects that are not discoverable due to the way AF handles URIs. You'd need to decide to use this from the outset (or not) in order to avoid this type of problem. Thus, it isn't a good candidate for a "flipper"
        • why not just have UUIDs for everything?
        • some folks use UUIDs for some things and NOIDs for others (e.g. when you need a pretty looking ID)
        • ...extended discussion about how NOIDs work...including:
          • where are the slowdowns happening? (db lock on noid minting)
          • how big are they? (~20 s per ingest)
        • also - fedora has the ability to wire in a module to mint ids, but it seems that nobody has done this in production (??)
        • if you know how many things you are going to ingest, you can mint a bunch of IDs ahead of time - maybe this is a change (nontrivial) that could be made to afnoid
        • cam156 offers an alternative approach that justin will take a stab at
  • Next week: