Repost: OpenAI Codex app: Vibe update from my old qqman R package to ggplot2 with plan+execute. | R bloggers

Reposted from the original on https://blog.stephenturner.us/p/openai-codex-app-qqman.

Earlier this week I reposted a Bluesky post from Ethan Mollick. I’m on the same page here: there are certainly problems, but these tools have irreversibly changed the practice of software development (and IMHO, on balance for the better).

Claude Code has been getting everyone excited lately (including me), which even goes so far as to make it into the popular press (see the Atlantic article on CC). Codex has been around for a while, but this seems like OpenAI’s attempt to challenge CC’s position.

This week OpenAI launched a special one Codex desktop app

positioned as a focused user interface for running multiple agents in parallel, keeping changes isolated via built-in work trees, and extending behavior with skills and scheduled automations. Like the demo I highlighted in my
previous post about Claude CodeCodex uses worker trees to isolate tasks/PRs as conflict resolution primitives, has a plan mode (/plan) to force prior parsing and queries, can use skills as reusable bundles that can connect to external services, and has automations for recurring background tasks.

Download it openai.com/codex. It is free to use for an indefinite period of time. Or works with your Plus/Enterprise/Edu/whatever account.

My most cited article is about some R code I wrote 15 years agoand later converted into a package, for making Manhattan plots from GWAS data. The article in JOSS has been cited ~1000 times, and the preprint another ~2100 times.

Turner, S. D. (2018). qqman: an R package for visualizing GWAS results using QQ and Manhattan plots. Journal of Open Source Software3(25), 731. https://doi.org/10.21105/joss.00731.

I wrote the original code with ggplot2 and then refactored everything to use basic R, because at the time, at least with the way I wrote the package, millions of points were plotted with ggplot2 painfully slowly.

I have received many feature requests for qqman, but unfortunately I don’t have time to respond to them. Many of these feature requests would be easy to implement if the package actually used ggplot2, since it’s been so long since I’ve done anything in basic R that I’ve forgotten how to do pretty much everything.

So I fired up the Codex app, put it in plan mode and asked for help. Along the way I am asked questions about my preferences.

Here is a screenshot of the plan (shortened).

#YOLO, execute the plan.

When I rebuilt the package, there were a few of those “no visible global binding for global variable…” comments that I had to fix, but otherwise everything worked. Everything uses ggplot2 now. The function returns a ggplot object. Column names are no longer so strict. It’s fast. It uses ditch to label points so they don’t overlap. Documentation has been updated. The README has been updated. The vignettes have been updated. It went off and did this in about 5 minutes.

Here is one of the plots that comes from this.

I’m not going to update the package because I still don’t have time to maintain it or respond to feature requests. But that shouldn’t stop You

if you want to do something with qqman or another open-source R/Python package that the package doesn’t currently do.

Related