2018-09-24: Development meeting

Kacper, Craig, Damian, Ian, Kyle, Mike, Tommy

Agenda

  • Updates
  • [36-month plan]
  • Discussion of Roadmap and Gantt
    • KC: Need to clarify Forking, may not be a priority
      • KK: Workspaces are a webdav folder with a given tale ID.
        • Anyone with access would be able to modify
      • CW: What about “Remix” as a short-term solution?
        • Can we implement remix without forking? MH/KK: Yes
        • KK: Remixing is just taking the initial parameters
      • MH: Forking was a different idea. Each time you open a tale, you get a local copy.
      • CW: If I remix, can I edit the code?
        • KK: Remixing is taking the data without the notebook
        • CW: Remixing is on 3 dimensions – code, data, environment
        • KK: We get 2 out of 3 with remix, the 3rd with forking
        • MH: Remixing is creating a tale with this data
      • KC: We have a good use case with Logan
        • ~5 models that do the same thing, wants to apply to different data
        • KK: Logan’s environment already contains the code
        • KC: Need to define the minimum for the 36 month review
        • MH: This was all related to publishing – the point is to get tales that can be re-opened and re-run
        • TT: I think the original use case was what happens when a user launches a tale from DataONE (it gets ‘forked’)
    • KC: Discussion of publishing
      • KC: Could we push somewhere else?
      • TT: We have other DataONE servers that aren’t production
        • Some time ago, we were told that this would never be published to DataONE production
        • This was revised to include publishing to DataONE production with the finalized serialization format

Updates

  • Kacper:
    • Reviewed DMS work and deployed in production
    • Development environment – needs issue
  • Craig:
    • Roadmap and AHM/Workshop output review.
    • Initial contact regarding WT specific carpentries.
  • Damian
    • Looked into problem positioning the modals (feedback from Tommy - https://github.com/whole-tale/dashboard/issues/271)
    • Added ability to launch tale from view page (https://github.com/whole-tale/dashboard/issues/211)
    • Looking at how to validate local storage state
    • Switched to one-week sprints, nothing for WT
    • Added new public Slack channel to discuss UI testing on Ember
  • Ian:
    • Started looking at roadmap, will provide feedback
  • Kyle:
    • Read roadmap document
    • Concerned about publication not being in 36 month release
    • Concerned about testing – Selenium is expensive
      • CW: NDS uses Protractor/Selenium. It’s worked, has been fragile, but seems worth the effort
      • KK: We at least need a smoke test
      • KK: The point of tests is to make sure things work during development and release.
    • Would be good to go into more detail on some things.
  • Mike:
    • Working on dynamically adding data to a session
    • Comment on UI testing
      • Used something similar to Selenium, which they found useful
  • Tommy:
    • Out last week
    • Working on persisting the publish modal (linked in smartsheet)
    • Would like to add ‘doi’ field to the Tale structure (https://github.com/whole-tale/girder_wholetale/issues/167)
    • Reviewed dashboard issues & closed/edited now non-relevant issues (either not applicable after the AHM or already completed)
    • To Do:
      • Create/configure a publishing branch
        • Eventually move my personal deployement over to probably-cattle
        • Make dev2 the publishing deployment after
      • Can we close https://github.com/whole-tale/whole-tale/issues/38 ?
        • CW: Please don’t until we get confirmation from PI/stakeholders on Workspaces
    • Discussion of development instances:
      • CW: Let’s try the compose environment first
      • KK: Can also use Nebula, which is unstable but available