R Expert — Idiomatic R Language Guide

R Expert — Idiomatic R Language Guide

This skill captures the non-inferable delta: R-version-pinned high-drift items, this project's toolchain preferences, and where to delegate. Generic tidyverse style (naming, spacing, assignment, basic pipes, vectorization, pre-allocation) is assumed — follow the tidyverse style guide; do not restate it here.

High-drift items (verify against your R version)

• Native pipe |> over magrittr %>% — base R, no import. Available since R 4.1, but the _ named-argument placeholder (x |> f(y = _)) needs R 4.2+ (R 4.3+ to use _ with extraction, e.g. _$col). Without _, the piped value can only fill the first argument.
• Lambda shorthand \(x) — backslash function syntax, R 4.1+ (e.g. purrr::map_dbl(x, \(d) d^2)).
• testthat 3rd edition semantics (expect_snapshot(), parallel tests, stricter expect_*) — opt in via DESCRIPTION's Config/testthat/edition: 3. See r-lib-testing.

Project toolchain preferences

• User-facing messages/errors via cli::cli_abort() / cli::cli_inform() / cli::cli_warn() (not bare stop() / warning() / message()). cli markup ({.val}, {.code}, pluralization) gives consistent, styled output.
• Format with styler; lint with lintr.
• Performance / IO stack: bench (benchmarking), data.table / arrow / vroom (large data + fast IO), future + furrr (embarrassingly parallel).

Delegation

• Package development → r-lib-package-dev, with deep dives in r-lib-testing (testthat), r-lib-cli (user-facing messages), and r-lib-lifecycle (deprecation / versioning).
• Shiny apps → shiny-bslib (layouts, components) and shiny-bslib-theming (theming, dark mode, brand.yml).
• Style-guide details (naming, spacing, pipes, vectorization): the tidyverse style guide — do not restate here.

References

• Tidyverse Style Guide — canonical style reference
• Advanced R (Hadley Wickham)
• R for Data Science (2e)