owned this note   changed 6 years ago  
Linked with GitHub

pyOpenSci Meeting Notes - 5 December 2019

Attendees

  • Leah Wasser (Earth Lab, University of Colorado - Boulder)
  • David Nicholson (Emory University)
  • Jenny Palomino (Earth Lab, University of Colorado - Boulder)
  • Martin Skarzynski (NIH)
  • Max Joseph (Earth Lab, CU Boulder)
  • Aaryn Olsson (University of Arizona Adjunct, Planet Labs)
  • Luiz Irber (DIB Lab, UC Davis)
  • Chris Holdgraf (UC Berkeley, Project Jupyter)
  • Sasha Kielbowicz
  • Sander van Rijn (Leiden University, Software submitter)

Agenda

  1. New Software Submission: https://github.com/pyOpenSci/software-review/issues/15 Can someone with some understanding of multi-fidelity analytical benchmark functions help us understand if we can review this package? he'd like to submit to JOSS as well. It would be nice if we can review.
    DESCRIPTION BELOW:

Analytical benchmark functions are a like well-established datasets in the field of machine-learning: by comparing your algorithm on the same ones, you can accurately compare your results. They are essential for reproducibility, as typos in the formulas will obviously give very different results, but are not an 'external' utility like the examples given under Package Categories in the guidebook.

Several of our collaborators are still not sure what this package this does.
This package may be out of scope for us right now.

We should consider – adding additional info to the submission review. maybe about what a readme contains. etc…

We should point the author to the current guidelines that contain recommendations for structuring a README: https://www.pyopensci.org/dev_guide/packaging/packaging_guide

1.a. Notes of short discussion with submitting author (Sander) who joined a bit later:

  • README/documentation was indeed still a work in progress.
  • Was indeed uncertain about whether it was in scope for pyopensci.
  • Given the limited pool of currently available authors, package will not be submitted to pyopensci for now, but just to JOSS instead. May resubmit when pyopensci has grown, if the package is definitively considered within scope.

How do we track people / reviewers, etc?

Do we want to keep track of people somehow that are willing to review ?

Google Sheet / Form –
Give people the option to opt in – or make it anonymous ??

3. AGU 2019

What is the scope of pyopensci / ropensci

Here's some info on research compendia: https://research-compendium.science/

Here's some rOpenSci content relevant to reproducibility and research compendia: https://ropensci.github.io/reproducibility-guide/sections/introduction/

Link from Aaryn: https://github.com/benmarwick/rrtools

  • Chris: if we accepted these compendiums, it would be a lot of work to review as the paper would need to be reviewed.
  • Aaryn: Clarify Dev Guide Aims and Scope wrt compendia

what data should we collect?

  • Why does doumentation matter

    • could we consider asking questions that are much more basic
    • where do you see issues??

  • Have you developed a python package?

    • if so - why?
      • did it include docuentation
      • tests,
      • CI
      • etc
    • if not - why?

  • Possibly more basic questions to start a group discussion at AGU session

    • How do you share your analysis code with others on your team, others, the broader scientifc community? Challenges there?
    • How do you find code examples for your work now? Challenges there?

Post AGU – IRB –

We could potentially team up with ropensci on the education side of things.

martin: there are mnay options associated with python packages

​​​​How to best manage an old school (prepared slides only) room??

Static link to a Dashboard page for attendees to explore, possibly with walkthrough, though attendees could be offline as well given the volume of folks accessing network

A few notes from convos with Leo and Lindsey and others

Action items - What could be done to make software “value” more visible – the academic challenge of getting “Credit”

Even the open source software – abstracts lose attention because people often get funding for research not software
Publication –- when you publish you don’t get the same credit as you do for papers
What are the challenges associated with creating reproducible code /tools
Have
What is missing from your toolkit to make workflows more reproducible?
What sorts of support do you need to…
List the components of a good open source software package
Which do you understand the least
Discovering/ finding tools…

  1. "Organizing Code" blog post
Select a repo