August 31, 2016

Time/Place

This meeting is a hybrid teleconference and IRC chat. Anyone is welcome to join. Here is the info:

• Time: 1:00pm Eastern Daylight Time US (UTC-4)
• Dial-in Number: (641) 715-3570
  • Participant Code: 304589#
  • International numbers: Conference Call Information
  • Web Access: https://www.freeconferencecallhd.com/wp-content/themes/responsive/flashphone/flash-phone.php
• IRC:
  • Join the #islandora chat room via Freenode Web IRC (enter a unique nick)
  • Or point your IRC client to #islandora on irc.freenode.net

Attendees

• Ed Fujikawa
• Jared Whiklo 🌟
• Diego Pino Navarro
• Bryan Brown
• Nick Ruest
• Daniel Lamb
• Aaron Coburn
• Mark Cooper
• Don Richards
• Kirsta Staplefeldt
• Kim Pham
• Nat Kathan
• Caden Armstrong
• Mark Matienzo
• Melissa Anez

Agenda

1. Messaging
2. Sprint updates
3. Zookeeper

• totally awesome
• Totally distributed
• Totally gonna make you install three servers :(

4. CLAW IR Roadmap

• Phase 0: Get requirements from IR community & formalize in a prospectus
• Phase 1: Re-implement 7.x-1.x IR features
• Phase 2: Implement new features suggested by community

5. Feel free to add agenda items...

Minutes

1. Messaging

   Be nice in all interactions online. We are all friends here.

   Diego: feels like he is being treated like he is stupid, and we are trying as a community. It is complex to stand up for the needs of our community when each time we get a "no no". The whole OWL conversation has been going on for months now. I think we should stay with PCDM and make PCDM better, but the whole issue thing is getting to personal.

   Nick: Mike (Giarlo) and I talked, there has been sniping on both sides. Which isn't good. Maybe tomorrow's PCDM call will help a lot when people are using voices.

   Danny: We feel that if we don't win this fight then we will be forced to adopt the Hydra Works model, and that isn't going to happen. We will try to work with the PCDM, and they have already made their decision while we are still at the decision making process.

2. Sprint Updates

   Virtual daily stand-ups, starting in the September sprints. Any time before 10am. What you working on, what you will be working on, what any blockers are. Just to keep the ball rolling

   • Diego: get the main islandora module, jsonld module with normalizer and serializer, a new storage class at once store in Mysql but also send to microservices. Created a new context, which lead to a new issue around uniform names.

   • Jared: Content Negotiation to use Accept header to set format.

   • Kim: Documentation versioning, proposed workflow using tagging. Trying to find a way to not have to create a folder each past release and using anchor tags to provide links.

   • Bryan: Homework assignments, learning a lot about W3 tech documents. JSON-LD. Has a Zotero library of stuff.

   • Melissa: Reading the specs, also if the MVP is hammered down. Would like to maybe spend some time in Drupal 8 from the front end perspective in the next sprint.

   • Nick: Did homework, involved in modelling discussions. This week is leading the Fedora Import/Export sprint (using BagIt)

   • Kirsta: PCDM in the MVP. Following all I can. Wondering if in issue 341 to have some sort of statement and whether PCDM will be part of the MVP.

   Danny: We want it to be in the PCDM 1.0 spec, but now they have made changes based on their experiences and want that to be the common understanding. We are not sure that we want to use that as our base logic. PCDM 2.0 is still a work in progress.

   Diego: Do we stop working on ORE.

   Danny: We are going to use this one-way or another. Focusing on a simple RDF / NonRDF entities setup and circle back later to see where we standing with PCDM.

   Danny: Sync and ZooKeeper, Alpaca microservices thing is more or less ready and with make a post for public critique. Did not read the IIIF spec. On track to have a list of possible MVP parts.

   Kirsta would like some clear direction on how Islandora CLAW will/won't be implementing PCDM. To help make arguments to the larger administration for more developer time. Would be useful to have posts on the Islandora mailing list.

   Nick: that's part of the purpose of this sprint; modelling. We can do PCDM 1.0 no problem. PCDM 2.0 is still up in the air, because it is being worked out as we speak. We can't model our stuff in PCDM 2.0 because it isn't finalized.

   Jared: Ideally PCDM discussions can/should be its own thing because the community around it may not involve CLAW or Hydra. In past e-mails sent to multiple mailing lists which resulted in discussions would lose one community from the list due to a mistake in replying.

   Aaron: Initially there was a desire to have discussions around PCDM posted to the Fedora mailing list. But if PCDM grows it is possible that someone will use the data model without using Fedora.

3. Zookeeper

   Danny worked through Sync. Full bi-directional replication can and will happen.

   However, once things go asynchronous and move into the background, you start to lose guarantees. ZooKeeper can help with that, it does a lot more. But you use it to do "locks", I can say hey I am working on resource A, no one touch it. I need that lock in several places (Fedora, Drupal). ZooKeeper is meant to be run in a cluster, there is a standalone dev version. But ideally you run 3 minimum. So be aware of that. Nothing stopping you from running just one.

   Aaron: You can deploy 3 instances on a single machine, but that is not ideal.

   Diego: Using ZooKeeper to do Solr sharding, it used 3 zookeepers on the same machine on 3 different ports.

   Danny: Want people to understand that this is a thing, and if you have concerns please voice them.

   Aaron: ZooKeeper has a pretty serious learning curve, and distributed systems have a serious learning curve. There are problems you might not have thought about and ZooKeeper solved these problems. I would choose to use ZooKeeper. There are other programs and languages that do this, but if you have Java/Camel already running then ZooKeeper is a good base.

   Danny: I have fallen into this line of thinking because of the difficulty in solving these complex problems in a "roll your own" solution. Apache Curator could sit on top of ZooKeeper and solve the exact problem we are working on.

   Nick: Traditionally we use vagrant to develop. Is this where we move to Docker?

   Danny: We can create a development setup.

   Jared: If it will run on a single machine with 3 ports, we can do that. Vagrant is not a production system, it is a development/demo tool.

   Jared: Is this a small application we we can stick it on other machines already running other services.

   Aaron: It is pretty light weight, in the CPU sense. It does more I/O though. Depends on your usage.

4. IR proposal

   Bryan: Has some place holder text, but wanted to get some community review of the direction.

   Phase 1: make sure CLAW can do everything that 7.x-1.x can do.

   Once we are past that, we can start creating documentation around migrating IR items. Once that is done, then we look to what new features we can implement.

   Danny: Getting requirements and developing a prospectus can be hard. But the process is sound. Ask for help if you are having trouble getting participation.

Scholar is a huge part of the 7.x-1.x stack, so something like this will be a vital part of the CLAW implementation.

Bryan: Gather up ontologies that seem well suited to scholarly publications.

Diego: Can I create issues related to Drupal 8 and IR concerns, to show how that might work and gather feedback on possible options.

Nick: Also we can look at Sufia and see how they are modeling their scholarly data.

Bryan: I would love to get a tour of a Hydra system to see how they are handling/modeling some of their scholarly work.

Aaron: We have a Hydra installation albeit an old one, and could show you some of the parts of it. There are probably more members of the Hydra community that have newer versions and would be more than happy to take you on a tour.