A few weeks ago I released (so far only on r-universe) the 0.1.0 version of my array computing framework for R (jit compiles, supports autodiff and runs on CPU and GPU): Check it out here: https://github.com/r-xla/anvil #rstats

@krzysztof_dyba I posted this based on the key dates given here: https://user2026.r-project.org/participation/keydates/ Unfortunately I also can't find the submission page ...

@useR_conf The website says submissions for #useR2026 started yesterday, but I can't find where to submit my abstract. Has this date been changed?

Submission of UseR 2026 talks starts today :-) #rstats

If anyone is interested in giving it a try: https://r-xla.github.io/anvil/articles/anvil.html It's still in the pre-0.1.0 phase, but it's already somewhat useable :) Any feedback would be much appreciated.

@nrennie Wow, this looks amazing! :)

Has anyone tried using S7 for defining custom error classes and can tell me how well this has worked? #rstats

@kupac the criticism I would have with R is that not everything is an array, including the representation of scalars as 0-dimensional arrays.

@eliocamp interesting observation! But with tab-autocomplete from AI assistants it becomes less of an issue.

@eliocamp In the R REPL you get some autocomplete for nested structures (like for `l <- list(a = list(b = 1))`, `l$a$` will autocomplete to `l$a$b`, but this only works by actually evaluating the partial expression `l$a` to know about the existing names.

@eliocamp Python is a dynamic language just like R (I guess this is changing as they moving into the JIT direction and all the type hinting; I don't really follow the Python ecosystem so I can't really say a lot here) so the type of a.something() is only known at runtime. Therefore, you don't get autocomplete. Think of it like doing a$b() in an R script, where you also don't get autocomplete.

Does anyone know a good professional designer for R package logos? #rstats

https://www.heise.de/en/news/Goodbye-Microsoft-Schleswig-Holstein-relies-on-Open-Source-and-saves-millions-11105459.html

Heya R devs - did you know you can run all your favorite GitHub actions on Codeberg?

Codeberg is rolling out Forgejo actions - an (almost) drop-in replacement for GitHub actions, which means we can (almost) use `r-lib/actions` directly on a free and open source platform!

Just a couple tweaks are needed, and for your convenience I'm automatically mirroring r-lib/actions and applying those changes so they're ready to use.

https://codeberg.org/r-codeberg/r-lib-actions

#rlang #rstats #codeberg

#rstats Jarl 0.2.0 is available!

Jarl is a very fast R linter, written in Rust.

New since the initial release:

* 10 new rules, including 5 opt-in rules for testthat that can be activated with:

- `--extend-select TESTTHAT` in the CLI
- `extend-select = ["TESTTHAT"]` in jarl.toml

* 3 CLI args were renamed

* bug fixes and new docs:
- vignette "Tutorial: add a new rule"
- new sections on how to use Jarl in Helix and Neovim

Changelog: https://jarl.etiennebacher.com/changelog

@dodecadron roxytests seems pretty nice! thanks.

@dodecadron yeah, something like this, but embedded in comments so there is no runtime overhead when running the code in a standard way.

Wouldn't it be nice if we could define tests within function bodies, so one could really make it clear if a test is testing a very specifc part of an implementation? #rstats

Has anyone else observed that updating the github actions checkout workflow to v6, breaks the pkgdown workflow and maybe even has a fix for it? #rstats

Name a better love story than R and 64-bit integers ... #rstats