• Runs specific code for test cases.
• testthat and RUnit packages
• usethis::use_testthat()

context("All my data tests")
test_that("Data is installed", {
    expect_is(sri24_image_df(), "data.frame")
})

• see expect_equal, expect_identical, etc.

• Calculates the code coverage: % of lines tested when package is tested
  • excludes white space
  • covr::package_coverage() - defaults to tests unit tests only
  • covr::package_coverage(type = "all") - runs tests, vignettes, examples.

Online services that allow you to view the coverage online:

• Coveralls: https://coveralls.io/
• Codecov: https://codecov.io/

• Check the Package: Build → Check Package.
  • Make sure Build → Configure Build Tools. Put --as-cran in Check Package area.

devtools::check(args = c('--as-cran'))

• Runs examples, tests, and check imports/suggests/etc.

devtools::check(args = c('--as-cran'))

Before submitting to CRAN:

• Run available::available("PACKAGE_NAME"): checks package name
• Shows notes: non-global visible things are a by-product of tidyverse non-standard evaluation

• Builds and checks R packages on Windows (Appveyor) and Linux/OS X (Travis CI)
• Sign up with GitHub, works well with GitHub
• Installs R, runs R CMD check --as-cran, and provides result

• usethis::use_travis() - creates .travis.yml
  • add warnings_are_errors: true

Add coverage: usethis::use_coverage() - pick coveralls/codecov:

after_success:
  - Rscript -e 'covr::coveralls(type = "all")'
  - Rscript -e 'covr::codecov(type = "all")'

• usethis::use_appveyor() - creates appveyor.yml

• Writing R Extensions - all things for a package

• CRAN policy (long): https://cran.r-project.org/web/packages/policies.html

• devtools::release() (needs NEWS.md and cran-comments.md)

• RStudio → Build → Build Source Package
  • builds vignettes, creates a .tar.gz, upload to https://cran.r-project.org/submit.html

• Clear out man folder
• Replace NAMESPACE with blank file with # Generated by roxygen2: do not edit by hand at the top
• Check your Imports/Suggests/Depends

• Use Imports. Only in very rare cases, use Depends, unless depending on R version.
• Use @importFrom directives instead of @import
  • helps you identify problems (moving functions)
• Use the “Go to file/function” search in RStudio
• Separate files for each function (personal preference)