Wednesday October 5th 2022: pyOpenSci: A diverse community that supports open tools that drive open science

Meta Pangeo Showcase
1

Announcement
Hello all, showcase will be taking a brief hiatus this week, but please still come to the community meeting if interested. Please join us at next weeks showcase to learn about pyOpenSci from executive director Leah Wasser!

Meeting Logistics!
Title: pyOpenSci: A diverse community that supports open tools that drive open science through peer review, mentorship and training; how can we support Pangeo?
Invited Speaker: Leah Wasser, Executive Director: pyOpenSci
When: Wednesday, October 5 2022 12PM EDT
Where: https://columbiauniversity.zoom.us/j/94877958106?pwd=UkE0UHF1U0x3VTVUNEJTam9mTXVHZz09</ins>
Abstract:
pyOpenSci builds a diverse community around the free and open Python tools that drive open science through an open peer review process. We also build technical skills needed to contribute to open source and that in turn support development of open science workflows through mentorship and training. pyOpenSci is now an independent organization fiscally sponsored by Community Initiatives and is funded by the Sloan Foundation. In that funding we outlined a model to support domain-specific communities such as Pangeo. We are excited to potentially collaborate with the Pangeo community. Come learn about the mission of pyOpenSci. We will discuss ways that pyOpenSci can support domain specific Python tools developed within the Pangeo community. We will also discuss ways that the Pangeo community can drive standards that pyOpenSci uses in our open peer review. We hope that the Pangeo community is excited as we are to support a diverse group of new and aspiring reviewers and open source contributors.
Relevant material: https://www.pyopensci.org, pyOpenSci · GitHub
Agenda:

  • 5-15 minutes - Community showcase
  • 5-15 minutes - Q&A / Community check-in
  • 20-35 minutes - Agenda and Open discussion

Kind regards,
The Pangeo Showcase Organizers

2

Your presentation was great @lwasser and the discussion about it was super thoughtful. I agree with @rabernat that pyOpenSci could have a big impact if it lessened the pain of managing dependencies. It’s a tough problem. A good start might be to not prescribe a list of dependencies for different domains (use only httpx for asynchronous web requests, not aiohttp, for example), but to advise projects and their developers to be more intentional about their choices. Require only one visualization package instead of two or three, and ideally the same one that related packages are using. Use only gdal or rasterio, and not both. I’m optimistic that this could limit the rate of growth of our dependency trees over time.

I appreciate that Pangeo makes these showcases open. I’ve wanted to participate more for a long time and in my current position it is much more possible to do so. I’m going to keep listening in.

2 Likes
3

Thanks Leah for your inspiring presentation and for your work on pyopensci. :pray:

And thanks @sgillies for your thoughtful comments about dependencies (and for all your open source work in general!) :pray: We’re very happy to welcome you to the community.

1 Like
4

Gosh - i’m sorry that I missed this post, @sgillies @rabernat i maybe didn’t have notifications turned on correctly. Sean, i’d be interested in talkin to you more about the dependency issue. i haven’t gotten to that yet in our packaging guide but will get there.

I have started talking with folks about things like locking envt files and pinning deps but haven’t gotten to best development practices. Perhaps when i get that I could ping you on that? i do want to get this right. i am going to be making some decisions around packaging in a space where there are many different opinions.