Reproducibility project: November 2024 report

In November, I have continued to work on exposing the columnar dependencies of operations.
Because this work builds on top of existing open pull requests, not much of it has made it to the pull request list, apart from #6988, #7024 and some further improvements to #6929. The rest of the changes is available in my GitHub fork, which I regularly rebase on top of master as pull requests get merged. Thanks to @abbe98 and @Sunil_Natraj, we merged some of my PRs on this topic (#6907, #6908), but the big backlog of open PRs remains a significant hurdle (as it increases the rebasing effort and increases the risk that I am building on top of changes that might not get accepted in the end). I am very open to suggestions on how to do this better: making smaller PRs, making bigger PRs, explaining the motivation of my changes better, merging my PRs without review after a certain delay… let me know!

Apart from that, I have been working on routine maintenance (responding to bug reports, PR review, cleanup PRs). I have also proposed @tfmorris to help out with the 3.9 release publication, but he intends to get it done on his side.

By default, I would continue work in this way, by accumulating PRs to be submitted when the existing ones get merged. In December, I intend to implement the interface to adapt columns mentioned in a workflow to the ones contained in the project, and start work on a pipeline-style representation of operations, following the designs established earlier this year.

2 Likes

I wish I could be more helpful with reviews, I'm afraid my expertise largely sits outside of that needed.