group picture of sprint attendees - including toddlers!

On July 18-19, we’ve had our first all-remote scikit-image sprint, as part of the scipy conference. Of course we missed seeing old friends and meeting new ones around the Scipy BBQ, but this remote sprint has been a great experience and brought many useful contributions to the project. Here are a few take-aways from this sprint.

  • All-remote means that all participants are on an equal footing: when sprinting at a conference, core devs tend to pay more attention to people in the same room than to people who are sprinting remotely, or to people they know better / who are more confident to ask questions, etc. Here all conversations were taking place on our Zulip chat forum, which made conversations more “distributed”, with everybody being able to ask questions or chime in on specific topics.

  • As usual, preparation is key. It was definitely useful to go through our list on open issues before the sprint and label some as “good first issue” or “sprint”. By “good first issue” we often think of issues adapted to contributors who are starting to contribute to open source in general. In a sprint, however, first-time contributors to your project may already have a lot of experience in other domains, and it’s worth labeling a few issues for this type of contributors. This is how we’ve been lucky to have a very experienced developer and new contributor writing our first Github action for an improved deployment of the documentation!

  • Regarding tools, zulip is very nice for chatting with its topic feature. We missed a bit not having a video platform with the possibility to create breakout rooms (although it was nice to have a common Google Meet room to be able to see other sprinters!). Other projects have had a great experience with the discord platform for sprinting, we’ll probably give it a try at the next sprint.

  • Even if you still need to make adjustments when sprinting on weekends, a remote sprint is more compatible with family-life as an in-person one. As you can see from our group picture, we even had three toddlers enjoying the sprint for a little while!

Here is a short summary of what we’ve been doing during the sprint and as follow-up work. These improvements will be available in our 0.18 release, or right now if you install the dev version.

  • new biology and medical-imaging datasets have been added to our data repo, opening the door to new tutorials for life science users.
  • we have added our first github action workflow for an improved deployment of the development documentation (#4843 and #4852)
  • bug fixes for the ransac algorithm (4844) and color string mapping in label2rgb (#4840)
  • automatic formatting of docstrings for improved consistency (#4849)
  • improved docstring for rgb2lab (#4839) and marching_cubes #4846
  • improved consistency of structure_tensor with the rest of the code base (#4841)
  • tutorial on visualizing 3D data (#4850)
  • default value for level parameter in find_contours (#4862)
  • ongoing work on a boundary tracing algorithm (#4853) and a faster convex-hull algorithm

Many thanks to all of our sprinters!

  • Alex de Siqueira
  • Clement Ng
  • Corey Harris
  • Emmanuelle Gouillart
  • Felipe Rodrigues
  • Gregory Lee
  • Joris Vankerschaver
  • Juan Nunez-Iglesias
  • Lars Grüter
  • Louis Maddox
  • Marianne Corvellec
  • Mathias Buissonier
  • Ruby Werman
  • Stéfan van der Walt
  • Volker Hilsenstein
  • Wendy Mak
  • Yogendra (@cedarfall)

As usual, a sprint is just how the story begins ;-)… we hope to see new contributors again on Github, and maybe become maintainers in the future!