2019-11-18: Development Meeting

Kacper, Mike L., Elias, Tim, Tommy, Mike H., Craig

Agenda

  • Updates
  • Discussion
    • Clean up old Tales? [name=Tommy]
      • Matt Jones tried to use one of his old Tales and things were a bit off (I pinged Kacper about this)
      • We might have some Tales that broke during upgrades and we might want to get rid of them (for example the loop 11 tale)
      • https://hackmd.io/aEGQcgVdSa2p_dfimkx77w
    • Dataverse/RO-Crate? [name=Tommy]
      • I noticed PDurbin forked my tale representation as RO-Crate
      • KK: my fault, don’t worry about it.
        • http://irclog.iq.harvard.edu/dataverse/2019-11-18#i_112752
        • http://irclog.iq.harvard.edu/dataverse/2019-11-18#i_112756
    • C2metadata [name=Craig / name=Tim]
      • The actual call notes
      • They want us to treat it like a blackbox and wire up WT to use it.
      • Can we think of C2Metadata as operating at one level that isn’t available from ptrace
        • We start with modeling in ptrace/system + YesWorkflow
    • RR next steps [name=Craig / name=Mike H.]
      • UI design - Kristina will present revised UI in ~2 weeks
      • Backend design -
        • Craig: Linking v Git branching (+ Girder?)
        • Mike’s design shows that you can easily compare between versions
        • Kacper: Make a document that enumerates what we want to do - e.g.,
          • Create a version
          • Compare file from version A with file in version B using my favorite tool
          • What’s in Mike’s UI…
        • Mike H: The number of things you can do on a filesystem is quite large
          • Designing something generic can be constrained (crippled) by too detailed use cases.
          • Filesystem seems to fit, is easily understood by someone dealing with a new system
          • Tentatively, implement versioning on the backend
            • Filesystem objects and metadata in Girder/Mongo
          • e.g., mapping metadata as a file

Updates

  • Tommy:
    • Currently reviewing some PRs for 0.9
    • Working on the README.html file for dataone package exports, so that user can run downloaded package locally
    • Pinged NCEAS people to test the analyze feature again
      • The goal is to demo this in production at AGU in a few weeks
      • It looks like the leadership team wants (At this point) more people using it on the test/demo sites
    • Will try to demo DataONE file hierarchy at the end of the week
  • Elias:
    • bdbag
      • Finished all tests and made necessary revisions to the implementation
      • Going to pull the newest bdbag code and do a rebase with my changes and submit a PR
  • Craig:
    • iSP pilot verification
    • C2Metadata planning
    • v0.9 PR review and (scope creep) planning
    • Reproducible run/versioning design and UI discussions
    • Addressing website update feedback (https://wholetale.xarthisius.xyz)
    • SOMOSPIE pilot tale (Taufer lab, UTK) discussion
  • Kacper
    • Zenodo publishing
    • Some of the things Craig mentions above ^^ (repro run, UI, C2M, website)
    • As mentioned optimized: x^4 + ax + b = 0 (mere 1000x speedup)
  • Tim:
    • C2Metadata meeting was interesting. Still a bit puzzled why SDTL is the wrong place to look for understanding what C2Metadata can say.
    • Clarification of cardinalities in model of system was informative.
  • Mike L.
    • Reproducible run/versioning design and UI discussions
    • Updated auth PR to include DataONE and OAuth cases
    • Containerized angular version of the dashboard but run into issues. Still work to be done.
  • Mike H.
    • Reproducible run/versioning design and UI discussions
    • Presented mockup for versioning