/PyOpenSci

Notes from the PyOpenSci BoF at SciPy 2018.

Notes from SciPy 2018

Comments and Motivation

Overarching scientific computing review

  • Community for OSS contributions
  • Technical mentorship
  • "Giving good advice on software is a skill."
  • Include code of conduct

Advantages for the Python community

  • reproducibility of code / experiments
  • "Does this actually make sense to anyone outside of the author?" / understanding
  • code pairing for reviewing and for creating software

Discoverability of packages in the scientific python ecosystem

  • Listing of scikit / scientific packages + quality of packages

Maintenance of scientific packages over time

  • response to issues and security risks
  • "Can you help me find homes for my open-source pets?" :)
  • Building a community of maintainers

Partnerships with journals

  • JOSS and Wiley
  • fast-tracking for publications (just based on paper, not code)

Checks for scientific software packages

  • Some manual; some personalized

Concerns

Has branding already been taken?

  • GitHub repo: not in active development
  • Twitter handle
  • Website

Questions

How much of the ROpenSci stack could be leveraged for Python?

  • Answer: almost of it!
  • Only a few language-specific components would be non-transferrable.
  • Estimated time commitment for porting over? TBD
  • Time commitment for organizers: half a day a week (Editor in Chief)

How to do recruiting for mentors?

  • average reviewer: 5 to 8 hours reviewing a package
  • cannot have the same people frequently (time constraints)

Initial funding: is it crucial for this kind of project?

  • rotating editor in Chief
  • 5 to 7 additional editors

Should PyOpenSci join ROpenSci?

  • Create new Github organization
  • Working collaboratively with the ROpenSci board
  • Create new board of Python editors / mentors / reviewers

What is the preservation policy of ROpenSci?

  • Same policy as JOSS

Would there be redundancy with JOSS?

  • ROpenSci = not in conflict with JOSS

Does it make more sense to have a joint community around scientific computing standards?

  • Putting a bunch of Python packages in the R stack doesn't make sense
  • But sharing community does :)

Next Steps

  • 1/2 page summary of collaboration with JOSS
  • Roadmap for timelines
  • Estimation of time commitments