R Packages Organize, Test, Document, and Share Your Code (Hadley Wickham, Jenny Bryan)（Z-Library）

【One-Line Pitch】 A practical, opinionated guide to turning loose R scripts into installable, tested, documented packages, written by the maintainers of the tidyverse toolchain. Best for R users who already write functions and want to share them reliably with others (or their future selves). 【Book Arc】 - **Opening (~0%–10%)**: Frames the whole game — what an R package actually is (code, docs, tests, data) and why packaging beats copy-pasting scripts. Introduces the core toolchain: devtools, usethis, testthat, roxygen2. - **Early (~10%–35%)**: Walks through creating a first package end-to-end (the `regexcite` example), covering `create_package()`, DESCRIPTION metadata, licensing, `check()`, `test()`, and `load_all()`. Solves the "where do I even start" problem. - **Middle (~35%–60%)**: Deepens structure and workflow — package states (source/bundled/binary), `.Rbuildignore`, dependencies and `use_package()`, exporting functions, and the discipline of running `R CMD check` iteratively. Also covers the "package within" refactor: extracting helpers from a messy script. - **Late (~60%–85%)**: The craft chapters — code style, respecting the R landscape, managing state with `withr`/`on.exit()`, package data, and the art and science of testing with testthat (basics, test design, advanced techniques). - **Ending (~85%–100%)**: Documentation with roxygen2, building a pkgdown website, and the release path — reverse dependency checks, CRAN submission, and handling failure modes. 【Key Takeaways】 - **Packages are the unit of reusable R code** (Opening): bundling functions, data, docs, and tests together is what makes code installable and dependable, not just shareable. - **Start with the whole game, then refine** (Early): the book's `regexcite` walkthrough gets a working package skeleton fast, then layers on correctness — a deliberate "make it work, then make it right" progression. - **`check()` is your constant health monitor** (Middle): run it often; ERRORs must be fixed, WARNINGs matter for CRAN, and NOTEs are worth eliminating when cheap. Long gaps between checks make bugs harder to isolate. - **Dependencies need explicit declaration** (Middle): even base-adjacent functions like `stats::median()` must be declared for CRAN — `use_package()` and the DESCRIPTION file are the mechanism. - **Testing is a discipline, not an afterthought** (Late): testthat's `test_that()`/`expect_equal()` patterns, plus `covr` for coverage, turn tests into a safety net that runs on every check. - **Respect the R landscape** (Late): code should avoid surprising side effects; use `withr` and `on.exit()` to manage and restore state cleanly. - **Documentation is part of the product** (Ending): roxygen2 comments generate function docs, and pkgdown turns them into a user-friendly website — documentation is written alongside code, not after. - **Releasing is a process with known failure modes** (Ending): reverse dependency checks, breaking-change awareness, and CRAN submission comments are all part of shipping responsibly. 【Reading Tips】 - **Deep-read the Early walkthrough** (~10%–35%): following `regexcite` step by step is the fastest way to internalize the workflow; skim it later as reference. - **Skim the contributor acknowledgments** (Early, ~16%–19%) — no conceptual content, safe to skip entirely. - **Treat the testing chapters (Late) as the hardest material**: design and advanced testing reward slow reading and hands-on practice with your own package. - **Keep `check()` running as you read**: the book's advice only sticks if you apply it to a real package you're building. - **Use the Ending as a checklist** when you're actually ready to release, not as front-to-back reading. 【Coverage Limits】 This guide is synthesized from stratified excerpts and a table of contents; specific code examples, chapter numbering, and detailed API behavior are only partially visible, so some finer points may be understated.