Cognitive Complexity

Cognitive Complexity is a measure of how difficult a unit of code is to intuitively understand. Unlike Cyclomatic Complexity, which determines how difficult your code will be to test, Cognitive Complexity tells you how difficult your code will be to read and comprehend.

A method's cognitive complexity is based on a few simple rules:

• Code is not considered more complex when it uses shorthand that the language provides for collapsing multiple statements into one
• Code is considered more complex for each "break in the linear flow of the code"
• Code is considered more complex when "flow breaking structures are nested"

Further reading

• Cognitive Complexity docs
• Cognitive Complexity: A new way of measuring understandability

A Nested Iterator occurs when a block contains another block.

Example

Given

class Duck
  class << self
    def duck_names
      %i!tick trick track!.each do |surname|
        %i!duck!.each do |last_name|
          puts "full name is #{surname} #{last_name}"
        end
      end
    end
  end
end

Reek would report the following warning:

test.rb -- 1 warning:
  [5]:Duck#duck_names contains iterators nested 2 deep (NestedIterators)

Feature Envy occurs when a code fragment references another object more often than it references itself, or when several clients do the same series of manipulations on a particular type of object.

Feature Envy reduces the code's ability to communicate intent: code that "belongs" on one class but which is located in another can be hard to find, and may upset the "System of Names" in the host class.

Feature Envy also affects the design's flexibility: A code fragment that is in the wrong class creates couplings that may not be natural within the application's domain, and creates a loss of cohesion in the unwilling host class.

Feature Envy often arises because it must manipulate other objects (usually its arguments) to get them into a useful form, and one force preventing them (the arguments) doing this themselves is that the common knowledge lives outside the arguments, or the arguments are of too basic a type to justify extending that type. Therefore there must be something which 'knows' about the contents or purposes of the arguments. That thing would have to be more than just a basic type, because the basic types are either containers which don't know about their contents, or they are single objects which can't capture their relationship with their fellows of the same type. So, this thing with the extra knowledge should be reified into a class, and the utility method will most likely belong there.

Example

Running Reek on:

class Warehouse
  def sale_price(item)
    (item.price - item.rebate) * @vat
  end
end

would report:

Warehouse#total_price refers to item more than self (FeatureEnvy)

since this:

(item.price - item.rebate)

belongs to the Item class, not the Warehouse.

A method with Too Many Statements is any method that has a large number of lines.

Too Many Statements warns about any method that has more than 5 statements. Reek's smell detector for Too Many Statements counts +1 for every simple statement in a method and +1 for every statement within a control structure (if, else, case, when, for, while, until, begin, rescue) but it doesn't count the control structure itself.

So the following method would score +6 in Reek's statement-counting algorithm:

def parse(arg, argv, &error)
  if !(val = arg) and (argv.empty? or /\A-/ =~ (val = argv[0]))
    return nil, block, nil                                         # +1
  end
  opt = (val = parse_arg(val, &error))[1]                          # +2
  val = conv_arg(*val)                                             # +3
  if opt and !arg
    argv.shift                                                     # +4
  else
    val[0] = nil                                                   # +5
  end
  val                                                              # +6
end

(You might argue that the two assigments within the first @if@ should count as statements, and that perhaps the nested assignment should count as +2.)

Duplication occurs when two fragments of code look nearly identical, or when two fragments of code have nearly identical effects at some conceptual level.

Reek implements a check for Duplicate Method Call.

Example

Here's a very much simplified and contrived example. The following method will report a warning:

def double_thing()
  @other.thing + @other.thing
end

One quick approach to silence Reek would be to refactor the code thus:

def double_thing()
  thing = @other.thing
  thing + thing
end

A slightly different approach would be to replace all calls of double_thing by calls to @other.double_thing:

class Other
  def double_thing()
    thing + thing
  end
end

The approach you take will depend on balancing other factors in your code.

Duplication occurs when two fragments of code look nearly identical, or when two fragments of code have nearly identical effects at some conceptual level.

Reek implements a check for Duplicate Method Call.

Example

Here's a very much simplified and contrived example. The following method will report a warning:

def double_thing()
  @other.thing + @other.thing
end

One quick approach to silence Reek would be to refactor the code thus:

def double_thing()
  thing = @other.thing
  thing + thing
end

A slightly different approach would be to replace all calls of double_thing by calls to @other.double_thing:

class Other
  def double_thing()
    thing + thing
  end
end

The approach you take will depend on balancing other factors in your code.

Classes should not assume that instance variables are set or present outside of the current class definition.

Good:

class Foo
  def initialize
    @bar = :foo
  end

  def foo?
    @bar == :foo
  end
end

Good as well:

class Foo
  def foo?
    bar == :foo
  end

  def bar
    @bar ||= :foo
  end
end

Bad:

class Foo
  def go_foo!
    @bar = :foo
  end

  def foo?
    @bar == :foo
  end
end

Example

Running Reek on:

class Dummy
  def test
    @ivar
  end
end

would report:

[1]:InstanceVariableAssumption: Dummy assumes too much for instance variable @ivar

Note that this example would trigger this smell warning as well:

class Parent
  def initialize(omg)
    @omg = omg
  end
end

class Child < Parent
  def foo
    @omg
  end
end

The way to address the smell warning is that you should create an attr_reader to use @omg in the subclass and not access @omg directly like this:

class Parent
  attr_reader :omg

  def initialize(omg)
    @omg = omg
  end
end

class Child < Parent
  def foo
    omg
  end
end

Directly accessing instance variables is considered a smell because it breaks encapsulation and makes it harder to reason about code.

If you don't want to expose those methods as public API just make them private like this:

class Parent
  def initialize(omg)
    @omg = omg
  end

  private
  attr_reader :omg
end

class Child < Parent
  def foo
    omg
  end
end

Current Support in Reek

An instance variable must:

• be set in the constructor
• or be accessed through a method with lazy initialization / memoization.

If not, Instance Variable Assumption will be reported.

Duplication occurs when two fragments of code look nearly identical, or when two fragments of code have nearly identical effects at some conceptual level.

Reek implements a check for Duplicate Method Call.

Example

Here's a very much simplified and contrived example. The following method will report a warning:

def double_thing()
  @other.thing + @other.thing
end

One quick approach to silence Reek would be to refactor the code thus:

def double_thing()
  thing = @other.thing
  thing + thing
end

A slightly different approach would be to replace all calls of double_thing by calls to @other.double_thing:

class Other
  def double_thing()
    thing + thing
  end
end

The approach you take will depend on balancing other factors in your code.

Classes should not assume that instance variables are set or present outside of the current class definition.

Good:

class Foo
  def initialize
    @bar = :foo
  end

  def foo?
    @bar == :foo
  end
end

Good as well:

class Foo
  def foo?
    bar == :foo
  end

  def bar
    @bar ||= :foo
  end
end

Bad:

class Foo
  def go_foo!
    @bar = :foo
  end

  def foo?
    @bar == :foo
  end
end

Example

Running Reek on:

class Dummy
  def test
    @ivar
  end
end

would report:

[1]:InstanceVariableAssumption: Dummy assumes too much for instance variable @ivar

Note that this example would trigger this smell warning as well:

class Parent
  def initialize(omg)
    @omg = omg
  end
end

class Child < Parent
  def foo
    @omg
  end
end

The way to address the smell warning is that you should create an attr_reader to use @omg in the subclass and not access @omg directly like this:

class Parent
  attr_reader :omg

  def initialize(omg)
    @omg = omg
  end
end

class Child < Parent
  def foo
    omg
  end
end

Directly accessing instance variables is considered a smell because it breaks encapsulation and makes it harder to reason about code.

If you don't want to expose those methods as public API just make them private like this:

class Parent
  def initialize(omg)
    @omg = omg
  end

  private
  attr_reader :omg
end

class Child < Parent
  def foo
    omg
  end
end

Current Support in Reek

An instance variable must:

• be set in the constructor
• or be accessed through a method with lazy initialization / memoization.

If not, Instance Variable Assumption will be reported.

Duplication occurs when two fragments of code look nearly identical, or when two fragments of code have nearly identical effects at some conceptual level.

Reek implements a check for Duplicate Method Call.

Example

Here's a very much simplified and contrived example. The following method will report a warning:

def double_thing()
  @other.thing + @other.thing
end

One quick approach to silence Reek would be to refactor the code thus:

def double_thing()
  thing = @other.thing
  thing + thing
end

A slightly different approach would be to replace all calls of double_thing by calls to @other.double_thing:

class Other
  def double_thing()
    thing + thing
  end
end

The approach you take will depend on balancing other factors in your code.

A NilCheck is a type check. Failures of NilCheck violate the "tell, don't ask" principle.

Additionally, type checks often mask bigger problems in your source code like not using OOP and / or polymorphism when you should.

Example

Given

class Klass
  def nil_checker(argument)
    if argument.nil?
      puts "argument isn't nil!"
    end
  end
end

Reek would emit the following warning:

test.rb -- 1 warning:
  [3]:Klass#nil_checker performs a nil-check. (NilCheck)

An Uncommunicative Variable Name is a variable name that doesn't communicate its intent well enough.

Poor names make it hard for the reader to build a mental picture of what's going on in the code. They can also be mis-interpreted; and they hurt the flow of reading, because the reader must slow down to interpret the names.