23.2 Checks

50 checks - for a full list, see the book.

The categories of checks are:

  • metadata

  • package structure

    • files are where they should be and there aren’t any non-standard top level files
    • it can be installed
    • package size isn’t > 5MB
    • in short: lots of pain that can be alleviated by using {usethis} for making any new files

  • DESCRIPTION

    • License: ____ must refer to a known license (https://svn.r-project.org/R/trunk/share/licenses/license.db) or must be a file LICENSE and the file must exist
    • All dependencies including Suggests must be installed
    • Every package listed in Depends must also be imported in the NAMESPACE or accessed with pkg::foo. If you don’t do this, your package will work when attached to the search path (with library(mypackage)) because it will also be attaching those packages that it depends on but will not work when only loaded (e.g. mypackage::foo()). In terms of good practice: don’t use Depends and always access other packages that your package depends on using pkg::foo.

  • NAMESPACE

  • R code

    • checks R scripts for errors - unlikely to get many issues here since they would have cropped up when running devtools::load_all() (which you probably would have before running a check)

    • can’t access functions from other packages which aren’t exported (:::)

      • you are not allowed to use ::: to access non-exported functions from other packages. Either ask the package maintainer to export the function you need, or write your own version of it using exported functions. Alternatively, if the licenses are compatible you can copy and paste the exported function into your own package. If you do this, remember to update Authors@R.

    • S3 generic/method consistency

    print <- function(x, ...){
  UseMethod("print")
}

# BAD
print.my_class <- function(x) {
  cat("Hi")
}

# GOOD
print.my_class <- function(x, ...) {
  cat("Hi")
}

# Also ok
print.my_class <- function(x, ..., my_arg = TRUE) {
  cat("Hi")
}
    • Don’t use assign() to modify objects in the global environment. If you need to maintain state across function calls, create your own environment with e <- new.env(parent = emptyenv()) and set and get values in it:
    e <- new.env(parent = emptyenv())

add_up <- function(x) {
  if (is.null(e$last_x)) {
    old <- 0
  } else {
    old <- e$last_x
  }

  new <- old + x
  e$last_x <- new
  new
}
add_up(10)
#> [1] 10
add_up(20)
#> [1] 30
    • Don’t use T or F in your functions - use TRUE and FALSE

  • Data

  • Documentation

    • checks for completeness in all documentation and all man/*.Rd files have correct syntax
    • checks maximum linewidth
    • checks examples in help files (can use devtools::run_examples() to check these alone)

  • Demos

  • Compiled Code

    • checks all foreign function calls with in compiled code

    • runs all tests (devtools::test())

      • if the devtools::test() passes but test within the R CMD CHECK fails, there’s something that’s changed in your testing environment and it can be tricky to figure out. I’ve found restarting R before running devtools::test() seems to make the problem apparent.

  • Vignettes

    • checks for dependencies in vignettes and that the files are in right place within the pkg directory

    • builds the vignettes

    • running vignettes requires that the package is installed (devtools::install())

    • can be pretty slow - if you’re wanting to quicky iterate through check()s and focus on other problems in your package:

      • add --no-build-vignettes to “Build Source Packages” field in project options
      • devtools::check(vignettes=FALSE)