2019-09-16: Development Meeting¶

Kacper, Tommy, Elias, Joe, Craig, Mike L

Agenda¶

Updates
Discussion
- Elias’ Agave BDBag integration design
  - Craig to take action to ask Kyle about bdbag config
- v0.8 status
  - Run button enabled while tale starting
  - Still need to check manual testing steps
  - KK: When launching a new instance and the image build fails, the error is not reflected in the instance status.
    - Error isn’t propagated from build image
- PI priorities
  - Publishing (to Zenodo) – in progress, UI/UX work pending
    - Planned for v0.9
    - Publish provider refactor?
    - Blocker: release of v0.8
  - Collaboration/sharing – need UI/UX input
    - This is next on Kristina’s priority list. Ideally, we’ll have UI by 9/26
    - Need to consider back-end changes
  - Improved Github integration
- Tales at Scale
  - Joe availability: Thursday before 3CT or Friday noon-1 CT or after 2:30pm CT.

Updates¶

Kacper:
- Worked on Dataverse r2d integration
- Met with Kristina, Mike L, Craig - UI/UX mostly related to 3rd party integrations
- Met with Joe, Mike H, Craig - Tale @ Scale
Tommy:
- I’m no longer sick \o/
- Working on ‘Run’ button bug.
  - I think I know what’s going on
  - In the view route, we return a Tale model here
  - One caveat is that we manually fetch the instance and attach it to the Tale model here before returning it
  - We have two cases:
    - New Tale
      - At this point the instance hasn’t been fully created so the status of the instance is 0
    - Existing Tale (’View’ is clicked from browse)
      - In this case the instance probably has a status other than 0 (enough time has gone by that the Tale has been fully created)
    - In both cases the instance on the Tale object doesn’t change from the initial state obtained in api-call.js
    - Example: If I redirect to the run page after a long period of time (90 seconds) where the instnace is fully created, the button has the correct state (because the instance attached to the Tale model is “fully matured”/instantiated)
  - Need a way to/need to update the instance on a status change
  - Should probably move this line elsewhere too (But I’m not sure if this is breaking anything)
Elias:
- Finished first run through of design doc for Tapis v2 bdbag integration.
- Currently working on Tapis v2 (Agave) integration
- Needing review: Add Agave Import Provider PR
  - Code has been reviewed by Kacper. Still needs to be test run. Also, waiting for master to be ready to merge new code.
Craig:
- Kubernetes helm chart
- Planning/design discussions
  - Tales at Scale
  - Provenance and repro task group. My current assertions:
    - We must preserve the image or equivalent
    - There is a value in generating information about what’s installed (human/machine readable)
      - Ideally restricted to what’s used by the code
    - We should create a UI for r2d config (can it be Jupyter compatible?)
    - We need something like the “Reproducible run” or record mode
  - Confirmed PI priorities:
    - Publishing (to Zenodo) – in progress, UI/UX work pending
    - Collaboration/sharing – need UI/UX input
    - Improved Github integration
- Prov/repro task group. My assertions:
  - We must preserve the image or equivalent
  - We need something like the “Reproducible run” or record mode
  - There is a value in generating information about what’s installed or was installed during the “reproducible run”. Ideally restricted to what’s used by the code
  - We should create a UI for r2d config (can it be Jupyter compatible?)
Mike L:
- Discussed UI/UX, changes for publishing to Zenodo and Dataverse
- Angular refactoring nearing completion
  - Cleaned up all tslint warnings
  - Added sticky footer
  - Restyling things to match brand colors (loading overlay, etc)
  - Next: restyle “Run” button component for multiple views (e.g. catalog and run view)
  - TODO: last few modals, page routes/titles, favicon, LOTS and lots of testing