first dplyr mondays

2 min read


dplyr mondays

I’ll be working on dplyr on at least mondays for the next few weeks. One day a week is a strong commitment, but having it set in stone in my agenda will help me commit to it.

dplyr is definitely the best project I’ve had the chance to work on professionally. I can’t help but feel somewhat ashamed about disappearing on it for no reason that I can fully explain myself. I know however that once I feel bad, it makes it much harder to come back or even simply reach out, especially online.

The success of dplyr is well established, and the project certainly has matured. Having a conversation about dplyr between the four authors at rstudio::conf was definitely one of my personnal highlights from the conference. This might be an oversimplification but in essence Hadley continued to lead, Lionel added tidy evaluation, and Kirill took over the various back ends, e.g. the data frame back end I initially worked on.

It turns out dplyr still needs me, so #dplyrmondays are an attempt to structure some of my time to commit (and hopefully 😆 push) to dplyr.

int i=0

Yesterday was the first iteration of #dplyrmondays. As much as I’m always eager to invent new things and discover uncharted territories, this is not something you do on day 0, so I’ve used my time to look at a few issues.

I opened one issue related to the BH 📦 and opened 4 ⬇️ requests.

One of them in particular led to flushing out an insidious bug that I first attributed to broom::augment, but then the 🏀 was bounced back to the stats::model.frame function. Thanks to Michael Chirico for promoting it to an R-devel thread, I only did the lazy thing and 🐦ed about it.

Four issues (and then some comments to other issues) feels minimal, but this includes getting back on track with the codebase and setup a more formal way of contributing, i.e. through pull requests, reviews, and systematic testing.

Today is another day, so I’ll work on another project. See you next monday.