Sharing Your Package

Licensing

• Just because your code is on GitHub doesn’t make it open source
  • The default is “All rights reserved”
• Choose a license that fits your intentions for the package
• CRAN requires an open source license for submission

Common licenses

Apply with usethis::use_*_license()

License	Type	Attribution required	Patent grant	Copyleft	Use function
MIT	Permissive	Yes	No	No	use_mit_license()
Apache-2.0	Permissive	Yes	Yes	No	use_apache_license()
GPL-3	Copyleft	Yes	Yes	Yes (strong/viral)	use_gpl_license()
LGPL-3	Weak copyleft	Yes	Yes	Yes (allows linking)	use_lgpl_license()
CC-BY	Creative Commons Attribution	Yes	No	No	use_ccby_license()
CC0	Public domain	No	No	No	use_cc0_license()

• MIT: Very short and simple; keep copyright & license notice.
• Apache-2.0: Explicitly grant patent rights to users and contributors. You can’t contribute code and then sue users for patent infringement.
• GPL-3: Requires derivative works to be distributed under GPL-3 (strong/viral copyleft).
• LGPL-3 (Lesser GPL v3): Allows linking from proprietary software; modifications to the library itself must be LGPL.
• CC0: Effectively places data in the public domain — good for datasets.
• CC-BY: Suitable for data packages where attribution is required; not share‑alike.

GitHub distribution

Installing from GitHub comes free with the development practices we have already discussed.

• Great for niche user bases who are comfortable with R
• No automatic update notifications (update.packages() won’t catch it)
• Need careful branching to avoid breaking users on main

R-Universe

• Service that builds and hosts R packages from GitHub repositories.

A good option for a collection of packages from your organisation.

• Builds binaries automatically
• Works with install.packages() via a custom repository URL
• No CRAN review process

R-Universe Documentation: https://docs.r-universe.dev

Setting up R-Universe

1. Create a GitHub repo named <username>.r-universe.dev

2. Add a packages.json registry file:

[
  {
    "package": "fishr",
    "url": "https://github.com/boshek/fishr",
    "branch": "*release"
  }
]

3. Install the r-universe GitHub App on your account

4. Wait ~1 hour - your universe appears at https://<username>.r-universe.dev

https://docs.r-universe.dev/publish/set-up.html

Installing from R-Universe

Users install your package like any CRAN package, but with an additional repository URL:

install.packages(
  "fishr",
  repos = c(
    boshek = "https://boshek.r-universe.dev",
    CRAN = "https://cloud.r-project.org"
  )
)

Or set it once in .Rprofile:

options(repos = c(
  boshek = "https://boshek.r-universe.dev",
  CRAN   = "https://cloud.r-project.org"
))

Then you can install with a simple install.packages()

install.packages("fishr")

CRAN

Where most R users expect to find packages.

• Builds binaries for Windows and macOS
• Updates tracked by update.packages()
• Distinct versioned releases
• Stringent review process

“If it hurts, do it more often”

Martin Fowler: https://martinfowler.com/bliki/FrequencyReducesDifficulty.html

Releasing to CRAN

usethis::use_release_issue()

• Good to use even if you’re not submitting to CRAN
  • helps you track the release process and ensures you don’t forget important steps.
• Edit/remove items that don’t apply (e.g., reverse dependency checks, blog post etc)

Can seem daunting, but the tooling makes it manageable. Very stringent requirements - but they exist to protect users.

Releasing to CRAN: TLDR

release()

• I use this instead of submit_cran()
• Runs through an additional list of pre-flight checks
• Builds the package bundle
• Submits to CRAN

This is a pre-flight checklist that complements your ongoing efforts to keep the package passing R CMD check cleanly.

The release process

1. Determine the release type (patch, minor, or major - dictates version number)
2. Update documentation: README.(R)md and NEWS.md
3. check() on multiple operating systems and R versions
4. Reverse dependency checks (if other packages depend on yours)
5. Submit to CRAN and wait for acceptance
6. Create a GitHub release and increment the version number for the next cycle

use_news_md()

use_news_md()

Adds a NEWS.md file to your package.

• Tracks version numbers
• Records user-facing changes between versions
• Displayed on CRAN and rendered by pkgdown

use_cran_comments()

use_cran_comments()

Creates cran-comments.md - submitted alongside your package:

## R CMD check results

0 errors | 0 warnings | 1 note

* This is a new release.

Notes: - Most times the template is sufficient - Note any CRAN check failures you are aware of and how you addressed them - There is always at least one NOTE on initial submission (“New submission”)

If your submission is rejected

Do not despair - it happens to everyone, including R-core and tidyverse developers.

• From the CRAN robot: fix the problem and resubmit
• From a human reviewer: do not reply to the email and do not argue - update cran-comments.md to explain how you addressed the issues, then resubmit