Code Smell 125 – ‘IS-A’ Relationship

We learned at school that inheritance represents an ‘is-a’ relationship. It is not.

TL:DR; Think about protocol and behavior, forget inheritance

Problems

Bad models
Unexpected behavior
Subclass overrides
Liskov substitution principle V…


This content originally appeared on DEV Community and was authored by Maxi Contieri

We learned at school that inheritance represents an 'is-a' relationship. It is not.

TL:DR; Think about protocol and behavior, forget inheritance

Problems

  • Bad models

  • Unexpected behavior

  • Subclass overrides

  • Liskov substitution principle Violation

Solutions

  1. Think in terms of behavior behaves-as-a

  2. Prefer composition over inheritance

  3. Subclassify always following 'behaves-as-a' relation

Context

IS-A relation comes from the data world.

We learned ERDs with structured design and data modeling.

Now, we need to think in terms of behavior.

Behavior is essential, data is accidental.

Sample Code

Wrong

// If you made Square derive from Rectangle, then a Square should be usable anywhere you expect a rectangle

#include <iostream>

Rectangle::Rectangle(const unsigned width, const unsigned height):
    m_width(width),
    m_height(height)
{
}

unsigned Rectangle::getWidth() const
{
    return m_width;
}

void Rectangle::setWidth(const unsigned width)
{
  /*Width and Height are independant*/
    m_width = width;
}

unsigned Rectangle::getHeight() const
{
    return m_height;
}

void Rectangle::setHeight(const unsigned height)
{
    m_height = height;
}

unsigned Rectangle::area() const
{
  /*Valid for both Rectangles and Squares*/
    return m_height * m_width;
}

Square::Square(const unsigned size)
    : Rectangle(size, size)
{
}

// OK for squares, bad for rectangles
// Protocol is bad, width and height are not relevant on squares
void Square::setWidth(const unsigned size)
{
    m_height = size;
    m_width = size;
}

// OK for squares, bad for rectangles
// Protocol is bad, width and height are not relevant on squares
void Square::setHeight(const unsigned size)
{
    m_height = size;
    m_width = size;
}

void process(Rectangle& r)
{
    unsigned h = 10;
    auto w = r.getWidth();
    r.setHeight(h);

    std::cout << "Expected area: " << (w*h) << ", got " << r.area() << "\n";
  //area is not well defined in squares
  //every square IS-A rectangle, but does not behave-like a rectangle
}

int main()
{
    Rectangle rectangle{3,4};
    Square square{5};
    process(rectangle);
    process(square);
}

Right

// If you made Square derive from Rectangle, then a Square should be usable anywhere you expect a rectangle

#include <iostream>

Rectangle::Rectangle(const unsigned width, const unsigned height):
    m_width(width),
    m_height(height)
{
}

unsigned Rectangle::getWidth() const
{
    return m_width;
}

void Rectangle::setWidth(const unsigned width)
{
  /*Width and Height are independant*/
    m_width = width;
}

unsigned Rectangle::getHeight() const
{
    return m_height;
}

void Rectangle::setHeight(const unsigned height)
{
    m_height = height;
}

unsigned Rectangle::area() const
{ 
    return m_height * m_width;
}

//No inheritance
Square::Square(const unsigned size):
    m_size(size)
{
}

unsigned Square::area() const
{ 
    return m_size * m_size;
}

void Square::setSize(const unsigned size)
{
    m_size = size; 
}

void process(Rectangle& r)
{
    unsigned h = 10;
    auto w = r.getWidth();
    r.setHeight(h);

    std::cout << "Expected area: " << (w*h) << ", got " << r.area() << "\n";
  //area is not well defined in squares
  //every square IS-A rectangle, but does not behave-like a rectangle
}

int main()
{
    Rectangle rectangle{3,4};
    Square square{5};
    process(rectangle);
    process(square);
}

Detection

[X] Manual

This is a semantic smell.

Tags

  • Inheritance

Conclusion

Real Number IS-A Complex number (according to math).

Integer IS-A Real number (according to math).

Real Number does not Behave-Like-A Complex number.

We cannot do real.setImaginaryPart() so it is not a Complex according to our Bijection

Relations

More Info

Credits

Photo by Joshua Rondeau on Unsplash

DRY - Don't Repeat Yourself - Every piece of knowledge must have a single, unambiguous, authoritative representation within a system.

Andy Hunt

This article is part of the CodeSmell Series.


This content originally appeared on DEV Community and was authored by Maxi Contieri


Print Share Comment Cite Upload Translate Updates
APA

Maxi Contieri | Sciencx (2022-03-26T14:58:18+00:00) Code Smell 125 – ‘IS-A’ Relationship. Retrieved from https://www.scien.cx/2022/03/26/code-smell-125-is-a-relationship/

MLA
" » Code Smell 125 – ‘IS-A’ Relationship." Maxi Contieri | Sciencx - Saturday March 26, 2022, https://www.scien.cx/2022/03/26/code-smell-125-is-a-relationship/
HARVARD
Maxi Contieri | Sciencx Saturday March 26, 2022 » Code Smell 125 – ‘IS-A’ Relationship., viewed ,<https://www.scien.cx/2022/03/26/code-smell-125-is-a-relationship/>
VANCOUVER
Maxi Contieri | Sciencx - » Code Smell 125 – ‘IS-A’ Relationship. [Internet]. [Accessed ]. Available from: https://www.scien.cx/2022/03/26/code-smell-125-is-a-relationship/
CHICAGO
" » Code Smell 125 – ‘IS-A’ Relationship." Maxi Contieri | Sciencx - Accessed . https://www.scien.cx/2022/03/26/code-smell-125-is-a-relationship/
IEEE
" » Code Smell 125 – ‘IS-A’ Relationship." Maxi Contieri | Sciencx [Online]. Available: https://www.scien.cx/2022/03/26/code-smell-125-is-a-relationship/. [Accessed: ]
rf:citation
» Code Smell 125 – ‘IS-A’ Relationship | Maxi Contieri | Sciencx | https://www.scien.cx/2022/03/26/code-smell-125-is-a-relationship/ |

Please log in to upload a file.




There are no updates yet.
Click the Upload button above to add an update.

You must be logged in to translate posts. Please log in or register.