Jenny Bryan

Was discussing pros and cons of LLM-generated code recently with colleagues and one big takeaway was this:

In humans, taste develops much faster than ability.

In coding assistants, ability develops much faster than taste.

📦 glue 📦 just updated to v1.8.0

Main improvement most #rstats
folks will notice is better docs!

1. Improved README / home page: https://glue.tidyverse.org/articles/glue.html
2. New "Get started" article thanks to tidy dev day contributors @stephhazlitt@fosstodon.org and Brennan Antone: https://glue.tidyverse.org/articles/glue.html
3. New article on wrapping glue, with techniques that transfer to rlang, cli, and withr: https://glue.tidyverse.org/articles/wrappers.html

@jameslairdsmith@fosstodon.org oh, yes, this advice has been there for years, but I think it bears repeating 😅 https://rstats.wtf/installing-packages.html#what-about-homebrew

@jrosell@mastodon.social I definitely use rig for installing R! Highly recommended! The reason for the toot is just the cumulative # of times I've encountered someone struggling with compiling packages from source. And they ended up in that situation by accident. Because they installed R with homebrew (and didn't specify the cask).

PSA: if you use Homebrew to install R, you almost always want to do:

brew install --cask r

instead of `brew install r`.

The cask approach installs the CRAN distribution of R, which will be compatible with CRAN-built binary packages and you will have a happier life 🍻 #rstats