Path: keywords.rb
Last Update: Thu Oct 22 23:00:56 -0700 2009

RDoc-style documentation for Ruby keywords (1.9.1).

David A. Black

June 29, 2009

Yes, I KNOW that they aren‘t methods. I‘ve just put them in that format to produce the familiar RDoc output. I‘ve been focusing on the content.

If anyone has a good idea for how to package and distribute it, let me know. I haven‘t really thought it through.

Also, if you spot any errors or significant omissions, let me know. Keep in mind that I‘m documenting the keywords themselves, not the entities they represent. Thus there is not full coverage of, say, what a class is, or how exceptions work.

Changes since first release:

  • Added __END__ (thanks Sven Fuchs)
  • Added ‘retry’ to retry example (thanks mathie)
  • Corrected description of whenrescue’ can be used (thanks Matt Neuburg)
  • Added else in rescue context (thanks Rob Biedenharn)


BEGIN   END   __ENCODING__   __END__   __FILE__   __LINE__   alias   and   begin   break   case   class   def   defined?   do   else   elsif   end   ensure   false   for   if   in   module   next   nil   not   or   redo   rescue   retry   return   self   super   then   true   undef   unless   until   when   while   yield  

Public Instance methods

Designates, via code block, code to be executed unconditionally before sequential execution of the program begins. Sometimes used to simulate forward references to methods.

   puts times_3(gets.to_i)

   BEGIN {
     def times_3(n)
       n * 3

Designates, via code block, code to be executed just prior to program termination.

   END { puts "Bye!" }

The current default encoding, as an Encoding instance.

Denotes the end of the regular source code section of a program file. Lines below __END__ will not be executed. Those lines will be available via the special filehandle DATA. The following code will print out two stanzas of personal information. Note that __END__ has to be flush left, and has to be the only thing on its line.

  DATA.each do |line|
    first, last, phone, email = line.split('|')
   puts <<-EOM
   First name: #{first}
   Last name:  #{last}
   Phone:      #{phone}
   Email:      #{email}

The name of the file currently being executed, including path relative to the directory where the application was started up (or the current directory, if it has been changed). The current file is, in some cases, different from the startup file for the running application, which is available in the global variable $0.

The line number, in the current source file, of the current line.

Creates an alias or duplicate method name for a given method. The original method continues to be accessible via the alias, even if it is overriden. Takes two method-name arguments (which can be represented by strings or symbols but can also be the bare names themselves).

  class Person
    def name=(name)
      puts "Naming your person #{name}!"
      @name = name

    alias full_name= name=

  p = = "David"        # Naming your person David!

  class Person
    def name=(name)
      puts "Please use full_name="

  p.full_name = "David"   # Please use fullname=

Boolean and operator. Differs from && in that and has lower precedence. In this example:

  puts "Hello" and "Goodbye"

the subexpression puts "Hello" is executed first, and returns nil. The whole expression thus reduces to:

  nil and "Goodbye"

which reduces to nil. In this example, however:

  puts "Hello" && "Goodbye"

the expression "Hello" && "Goodbye" is used as the argument to puts. This expression evaluates to "Goodbye"; therefore, the whole statement prints "Goodbye".

Together with end, delimits what is commonly called a "begin" block (to distinguish it from the Proc type of code block). A "begin" block allows the use of while and until in modifier position with multi-line statements:

    i += 1
    puts i
  end until i == 10

"Begin" blocks also serve to scope exception raising and rescue operations. See rescue for examples. A "begin" block can have an else clause, which serves no purpose (and generates a warning) unless there‘s also a rescue clause, in which case the else clause is executed when no exception is raised.

Causes unconditional termination of a code block or while or until block, with control transfered to the line after the block. If given an argument, returns that argument as the value of the terminated block.

  result ="lines.txt") do |fh|
    fh.each do |line|
      break line if my_regex.match(line)

The case statement operator. Case statements consist of an optional condition, which is in the position of an argument to case, and zero or more when clauses. The first when clause to match the condition (or to evaluate to Boolean truth, if the condition is null) "wins", and its code stanza is executed. The value of the case statement is the value of the successful when clause, or nil if there is no such clause.

A case statement can end with an else clause. Each when statement can have multiple candidate values, separated by commas.

  case x
  when 1,2,3
    puts "1, 2, or 3"
  when 10
    puts "10"
    puts "Some other number"

Case equality (success by a when candidate) is determined by the case-equality or "threequal" operator, ===. The above example is equivalent to:

  if 1 === x or 2 === x or 3 === x
    puts "1, 2, or 3"
  elsif 10 === x
    puts "10"
    puts "Some other number"

=== is typically overriden by classes to reflect meaningful case-statement behavior; for example, /abc/ === "string" checks for a pattern match from the string.

Opens a class definition block. Takes either a constant name or an expression of the form << object. In the latter case, opens a definition block for the singleton class of object.

Classes may be opened more than once, and methods and constants added during those subsequent openings. class blocks have their own local scope; local variables in scope already are not visible inside the block, and variables created inside the block do not survive the block.

   class Person
     def name=(name)
       @name = name

   david =
   class << david
     def name=(name)
       if name == "David"
         @name = name
         puts "Please don't name me other than David!"
   end = "Joe" # Please don't name me other than David!"
   joe = = "Joe"

Inside a class block, self is set to the class object whose block it is. Thus it‘s possible to write class methods (i.e., singleton methods on class objects) by referring to self:

  class Person
    def self.species
      "Homo sapiens"

Paired with a terminating end, constitutes a method definition. Starts a new local scope; local variables in existence when the def block is entered are not in scope in the block, and local variables created in the block do not survive beyond the block.

def can be used either with or without a specific object:

  • def method_name
  • def object.singleton_method_name

The parameter list comes after the method name, and can (and usually is) wrapped in parentheses.

defined? expression tests whether or not expression refers to anything recognizable (literal object, local variable that has been initialized, method name visible from the current scope, etc.). The return value is nil if the expression cannot be resolved. Otherwise, the return value provides information about the expression.

Note that the expression is not executed.

  p defined?(def x; end)   # "expression"
  x                        # error: undefined method or variable

  p defined?(@x=1)         # "assignment"
  p @x                     # nil

Assignment to a local variable will, however, have the usually result of initializing the variable to nil by virtue of the assignment expression itself:

  p defined?(x=1)          # assignment
  p x                      # nil

In most cases, the argument to defined? will be a single identifier:

  def x; end
  p defined?(x)            # "method"

Paired with end, can delimit a code block:

  array.each do |element|
    puts element * 10

In this context, do/end is equivalent to curly braces, except that curly braces have higher precedence. In this example:

  puts [1,2,3].map {|x| x * 10 }

the code block binds to map; thus the output is:


In this version, however:

  puts [1,2,3].map do |x| x * 10 end

the code is interpreted as puts([1,2,3].map) do |x| x * 10 end. Since puts doesn‘t take a block, the block is ignored and the statement prints the value of the blockless [1,2,3].map (which returns an Enumerator).

do can also (optionally) appear at the end of a for/in statement. (See for for an example.)

The else keyword denotes a final conditional branch. It appears in connection with if, unless, and case, and rescue. (In the case of rescue, the else branch is executed if no exception is raised.) The else clause is always the last branch in the entire statement, except in the case of rescue where it can be followed by an ensure clause.

Introduces a branch in a conditional (if or unless) statement. Such a statement can contain any number of elsif branches, including zero.

See if for examples.

Marks the end of a while, until, begin, if, def, class, or other keyword-based, block-based construct.

Marks the final, optional clause of a begin/end block, generally in cases where the block also contains a rescue clause. The code in the ensure clause is guaranteed to be executed, whether control flows to the rescue block or not.

  rescue ZeroDivisionError
    puts "Can't do that!"
    puts "That was fun!"


  Can't do that!
 That was fun!

If the statement 1/0 is changed to something harmless, like 1/1, the rescue clause will not be executed but the ensure clause still will.

false denotes a special object, the sole instance of FalseClass. false and nil are the only objects that evaluate to Boolean falsehood in Ruby (informally, that cause an if condition to fail.)

A loop constructor, used with in:

  for a in [1,2,3,4,5] do
    puts a * 10

for is generally considered less idiomatic than each; indeed, for calls each, and is thus essentially a wrapper around it.

  obj =
  def obj.each
    yield 1; yield 2
  for a in obj
    puts a



The do keyword may optionally appear at the end of the for expression:

  for a in array do
    # etc.

Ruby‘s basic conditional statement constructor. if evaluates its argument and branches on the result. Additional branches can be added to an if statement with else and elsif.

  if m.score > n.score
    puts "m wins!"
  elsif n.score > m.score
    puts "n wins!"
    puts "Tie!"

An if statement can have more than one elsif clause (or none), but can only have one else clause (or none). The else clause must come at the end of the entire statement.

if can also be used in modifier position:

  puts "You lose" if y.score < 10

then may optionally follow an if condition:

  if y.score.nil? then
    puts "Have you even played the game?"

See for.

Opens a module definition block. Takes a constant (the name of the module) as its argument. The definition block starts a new local scope; existing variables are not visible inside the block, and local variables created in the block do not survive the end of the block.

Inside the module definition, self is set to the module object itself.

Bumps an iterator, or a while or until block,to the next iteration, unconditionally and without executing whatever may remain of the block.

  [0,1,2,3,4].each do |n|
    next unless n > 2
    puts "Big number: #{n}"


  Big number: 3
  Big number: 4

next is typically used in cases like iterating through a list of files and taking action (or not) depending on the filename.

next can take a value, which will be the value returned for the current iteration of the block.

  sizes = [0,1,2,3,4].map do |n|
    next("big") if n > 2
    puts "Small number detected!"

  p sizes


  Small number detected!
  Small number detected!
  Small number detected!
  ["small", "small", "small", "big", "big"]

A special "non-object". nil is, in fact, an object (the sole instance of NilClass), but connotes absence and indeterminacy. nil and false are the only two objects in Ruby that have Boolean falsehood (informally, that cause an if condition to fail).

nil serves as the default value for uninitialized array elements and hash values (unless the default is overridden).

Boolean negation.

  not true    # false
  not 10      # false
  not false   # true

Similar in effect to the negating bang (!), but has lower precedence:

  not 3 == 4  # true; interpreted as not (3 == 4)
  !3 == 4     # false; interpreted as (!3) == 4, i.e., false == 4

(The unary ! also differs in that it can be overridden.)

Boolean or. Differs from || in that or has lower precedence. This code:

  puts "Hi" or "Bye"

is interpreted as (puts "Hi") or "Bye". Since puts "Hi" reduces to nil, the whole expression reduces to nil or "Bye" which evaluates to "Bye". (The side-effect printing of "Hi" does take place.)

This code, however:

  puts "Hi" || "Bye"

is interpreted as puts("Hi" || "Bye"), which reduces to puts "Hi" (since "Hi" || "Bye" evaluates to "Hi").

Causes unconditional re-execution of a code block, with the same parameter bindings as the current execution.

Designates an exception-handling clause. Can occur either inside a begin<code>/<code>end block, inside a method definition (which implies begin), or in modifier position (at the end of a statement).

By default, rescue only intercepts StandardError and its descendants, but you can specify which exceptions you want handled, as arguments. (This technique does not work when rescue is in statement-modifier position.) Moreover, you can have more than one rescue clause, allowing for fine-grained handling of different exceptions.

In a method (note that raise with no argument, in a rescue clause, re-raises the exception that‘s being handled):

  def file_reverser(file) {|fh| puts fh.readlines.reverse }
  rescue Errno::ENOENT
    log "Tried to open non-existent file #{file}"

In a begin/end block:

  rescue ZeroDivisionError
    puts "No way"

In statement-modifier position:

  while true
  end rescue nil

  david = Person.find(n) rescue

rescue (except in statement-modifier position) also takes a special argument in the following form:

  rescue => e

which will assign the given local variable to the exception object, which can then be examined inside the rescue clause.

Inside a rescue clause, retry causes Ruby to return to the top of the enclosing code (the begin keyword, or top of method or block) and try executing the code again.

  a = 0
  rescue ZeroDivisionError => e
    puts e.message
    puts "Let's try that again..."
    a = 1
  puts "That's better!"

Inside a method definition, executes the ensure clause, if present, and then returns control to the context of the method call. Takes an optional argument (defaulting to nil), which serves as the return value of the method. Multiple values in argument position will be returned in an array.

  def three
    return 3
    puts "Enjoy the 3!"

  a = three    # Enjoy the 3!
  puts a   # 3

Inside a code block, the behavior of return depends on whether or not the block constitutes the body of a regular Proc object or a lambda-style Proc object. In the case of a lambda, return causes execution of the block to terminate. In the case of a regular Proc, return attempts to return from the enclosing method. If there is no enclosing method, it‘s an error.

  ruby -e ' {return}.call'
    => -e:1:in `block in <main>': unexpected return (LocalJumpError)

  ruby19 -e 'p lambda {return 3}.call'
    => 3

self is the "current object" and the default receiver of messages (method calls) for which no explicit receiver is specified. Which object plays the role of self depends on the context.

  • In a method, the object on which the method was called is self
  • In a class or module definition (but outside of any method definition contained therein), self is the class or module object being defined.
  • In a code block associated with a call to class_eval (aka module_eval), self is the class (or module) on which the method was called.
  • In a block associated with a call to instance_eval or instance_exec, self is the object on which the method was called.

self automatically receives message that don‘t have an explicit receiver:

  class String
    def upcase_and_reverse

In this method definition, the message upcase goes to self, which is whatever string calls the method.

Called from a method, searches along the method lookup path (the classes and modules available to the current object) for the next method of the same name as the one being executed. Such method, if present, may be defined in the superclass of the object‘s class, but may also be defined in the superclass‘s superclass or any class on the upward path, as well as any module mixed in to any of those classes.

  module Vehicular
    def move_forward(n)
      @position += n

  class Vehicle
    include Vehicular  # Adds Vehicular to the lookup path

  class Car < Vehicle
    def move_forward(n)
      puts "Vrooom!"
      super            # Calls Vehicular#move_forward

Called with no arguments and no empty argument list, super calls the appropriate method with the same arguments, and the same code block, as those used to call the current method. Called with an argument list or arguments, it calls the appropriate methods with exactly the specified arguments (including none, in the case of an empty argument list indicated by empty parentheses).

Optional component of conditional statements (if, unless, when). Never mandatory, but allows for one-line conditionals without semi-colons. The following two statements are equivalent:

  if a > b; puts "a wins!" end
  if a > b then puts "a wins!" end

See if for more examples.

The sole instance of the special class TrueClass. true encapsulates Boolean truth; however, <emph>all</emph> objects in Ruby are true in the Boolean sense (informally, they cause an if test to succeed), with the exceptions of false and nil.

Because Ruby regards most objects (and therefore most expressions) as "true", it is not always necessary to return true from a method to force a condition to succeed. However, it‘s good practice to do so, as it makes the intention clear.

Undefines a given method, for the class or module in which it‘s called. If the method is defined higher up in the lookup path (such as by a superclass), it can still be called by instances classes higher up.

  class C
    def m
  class D < C
  class E < D

  class D
    undef m
  end   # Hi   # error   # error

Note that the argument to undef is a method name, not a symbol or string.

The negative equivalent of if.

  unless y.score > 10
    puts "Sorry; you needed 10 points to win."

See if.

The inverse of while: executes code until a given condition is true, i.e., while it is not true. The semantics are the same as those of while; see while.

while takes a condition argument, and executes the code that follows (up to a matching end delimiter) while the condition is true.

  i = 0
  while i < 10
    i += 1

The value of the whole while statement is the value of the last expression evaluated the last time through the code. If the code is not executed (because the condition is false at the beginning of the operation), the while statement evaluates to nil.

while can also appear in modifier position, either in a single-line statement or in a multi-line statement using a begin/end block. In the one-line case:

i = 0 i += 1 while i < 10

the leading code is not executed at all if the condition is false at the start. However, in the "begin"-block case:

  i = 0
    i += 1
    puts i
  end while i < 10

the block will be executed at least once, before the condition is tested the first time.

Called from inside a method body, yields control to the code block (if any) supplied as part of the method call. If no code block has been supplied, calling yield raises an exception.

yield can take an argument; any values thus yielded are bound to the block‘s parameters. The value of a call to yield is the value of the executed code block.