Learning Objectives

Style Guides

File Names

Syntax

Names

  • Only use lower-case snake_case.

    • Good

      red_apple
    • Bad

      Red_apple
      red.apple
      redApple
      RedApple
  • Variables should be nouns and functions should be verbs

  • Never use single letters as variables/functions

    • Good:

      num_sim <- 10
    • Bad

      simulate <- 10 ## verb
      x <- 10 ## single letter
  • Exceptions: Some letters are standard. Such as n for the sample size in rnorm(), runif(), etc…

Commas

  • Always put a space after a comma, not before (like English).
    • Good:

      mat[1, ]
    • Bad:

      mat[1 ,]
      mat[1 , ]
      mat[1,]

Parentheses

  • Don’t put a space in or around parentheses for functions.
    • Good:

      mean(x)
    • Bad:

      mean (x)
      mean(x )
  • Put spaces around parentheses for if statements, and for and while loops.
    • Good:

      if (x) {
      
      }
    • Bad:

      if(x){
      
      }
  • Put a space only after () for function creations.
    • Good:

      sim <- function(x) {
      }
    • Bad:

      sim <- function (x) {
      }
      
      sim <- function(x){
      }

Curley Braces

  • Whenever you use curly braces {}, the opening brace should be the last character on a line, and the closing brace should be the first character on a line.
    • Good:

      if (condition) {
        dostuff()
      }
    • Bad

      if (condition)
      {
        dostuff()
      }
      
      if (condition) {
        dostuff() }

if-else

  • else statements should be on the same line as a closing brace.

    • Good:

      if (condition) {
      
      } else if (condition2) {
      
      } else {
      
      }
  • Only use ifelse() where vectorization is important. If condition should be length 1, then use full if-else statements.

  • In a if-then statement, use || or &&, not | or &, since the latter two vectorize operations.

Infix Characters

  • An infix operator is one where arguments on both sides of it are used in a function. The alternative is prefix notation. Compare

    5 + 10 ## infix notation
    ## [1] 15
    `+`(5, 10) ## prefix notation
    ## [1] 15
  • Put spaces around all infix characters ==, +, -, *, /, ^, %>%, etc…

    • Good:

      x + 10
    • Bad:

      x+10
      x+ 10
      x +10
  • Exceptions: ::, :::, $, @, [, [[, unary -, unary +, :, and ?.

    • E.g. do ggplot2::qplot() or -1, not ggplot2 :: qplot() and - 1

Code Length

  • No lines should be greater than 80 characters.

  • To get a vertical line displaying the code length, in R studio go to “Tools > Global Options… > Code > Display”. Make sure “Show margin” is checked with “80” in the text box.

  • If a function call/definition is too long, break up arguments on new line.

    this <- is_a_very_long_function_call(
      that = "is",
      broken = "up",
      into = "many",
      indented = "lines",
      that = "are",
      easier = "to",
      read = NULL
    )

Other things

  • Always use <- for assignment, not =.

  • Always use " for strings, not '.

  • Always use TRUE or FALSE, not T or F

    • T and F are aliases for TRUE and FALSE, and so may be overwritten by the user, which is scary.
  • Don’t include non-ASCII characters in your code.

    • ASCII characters are lower case letters (a through z), upper case letters (A through Z), digits (0 through 9), and common punctuation.

    • Including non-ASCII characters will give you a CRAN note.

    • Non-ASCII characters usually show up when you copy and paste from the web. E.g. the following look normal but are non-ASCII (and are all different):

      • En Dash: “–”
      • Em Dash: “—”
      • Horizontal Bar: “―”
      • En Quad: “ ”
      • Em Quad: “ ”
      • En Space: “ ”
      • Em Space: “ ”
    • If you accidentally include such characters, you can find them with

      tools::showNonASCIIfile()

Functions

Function Argument Length

  • If you have a lot of arguments, indent the arguments on new lines.

    run_me <- function(this,
                       is,
                       a = "lot",
                       of = "arguments",
                       that = "are longer than 80 characters") {
    }

Function Length

  • You should break up your functions into discrete tasks.
    • Reduces duplicating code, so less prone to bugs.
    • Allows you to think more modularly about tasks, which makes code easier to reason about.
    • Makes it easier to combine code in new ways.
  • To force you to do this, make all functions be less than 50 lines. This is what Bioconductor does.

Explicit returns

  • In R, the last value evaluated in a function will be implicitly returned. I think this is bad practice since it makes it harder to reason about what R is returning. So always include a return() statement. Never do

    add_two <- function(x, y) {
      x + y
    }

    Always do

    add_two <- function(x, y) {
      return(x + y)
    }

Importing

  • Never use the @import tag in a package to bring all of a package’s exported functions into the NAMESPACE. This creates too much risk for name collision.

  • In a package, never import functions, always type the package where the function came from. This makes it easier to reason about namespaces. Never do

    #' @importFrom ggplot2 qplot
    plot_red <- function(x, y) {
      qplot(x, y, color = I("red"))
    }

    Always do

    plot_red <- function(x, y) {
      ggplot2::qplot(x, y, color = I("red"))
    }
  • Exceptions:

    • You will have to import infix functions (surrounded by percent signs). Such as

      #' @importFrom magrittr %>%
      #' @importFrom foreach %dopar%
    • There is a small performance penalty for using :: (about 5 µs). So import a function if you are iterating it \(\sim\) million times, and each iteration takes on the order of 1 ns.

Order of Arguments

  • Always place arguments with defaults after arguments without defaults.

  • Good:

    function(arg1, arg2, arg3 = NULL) {
    
    }
  • Bad:

    function(arg1, arg3 = NULL, arg2) {
    
    }

lintr


National Science Foundation Logo American University Logo Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.