8.4 — Access functions and encapsulation

Why make member variables private?

In the previous lesson, we mentioned that class member variables are typically made private. Developers who are learning about object-oriented programming often have a hard time understanding why you’d want to do this. To answer that question, let’s start with an analogy.

In modern life, we have access to many electronic devices. Your TV has a remote control that you can use to turn the TV on/off. You drive a car to work. You take a picture on your digital camera. All three of these things use a common pattern: They provide a simple interface for you to use (a button, a steering wheel, etc…) to perform an action. However, how these devices actually operate is hidden away from you. When you press the button on your remote control, you don’t need to know what it’s doing to communicate with your TV. When you press the gas pedal on your car, you don’t need to know how the combustion engine makes the wheels turn. When you take a picture, you don’t need to know how the sensors gather light into a pixellated image. This separation of interface and implementation is extremely useful because it allows us to use objects without understanding how they work. This vastly reduces the complexity of using these objects, and increases the number of objects we’re capable of interacting with.

For similar reasons, the separation of implementation and interface is useful in programming.


In object-oriented programming, Encapsulation (also called information hiding) is the process of keeping the details about how an object is implemented hidden away from users of the object. Instead, users of the object access the object through a public interface. In this way, users are able to use the object without having to understand how it is implemented.

In C++, we implement encapsulation via access specifiers. Typically, all member variables of the class are made private (hiding the implementation details), and most member functions are made public (exposing an interface for the user). Although requiring users of the class to use the public interface may seem more burdensome than providing public access to the member variables directly, doing so actually provides a large number of useful benefits that help encourage class re-usability and maintainability.

Note: The word encapsulation is also sometimes used to refer to the packaging of data and functions that work on that data together. We prefer to just call that object-oriented programming.

Benefit: encapsulated classes are easier to use and reduce the complexity of your programs

With a fully encapsulated class, you only need to know what member functions are publicly available to use the class, what arguments they take, and what values they return. It doesn’t matter how the class was implemented internally. For example, a class holding a list of names could have been implemented using a dynamic array of C-style strings, std::array, std::vector, std::map, std::list, or one of many other data structures. In order to use the class, you don’t need to know (or care) which. This dramatically reduces the complexity of your programs, and also reduces mistakes. More than any other reason, this is the key advantage of encapsulation.

All of the classes in the C++ standard library are encapsulated. Imagine how much more complicated C++ would be if you had to understand how std::string, std::vector, or std::cout were implemented in order to use them!

Benefit: encapsulated classes help protect your data and prevent misuse

Global variables are dangerous because you don’t have strict control over who has access to the global variable, or how they use it. Classes with public members suffer from the same problem, just on a smaller scale.

For example, let’s say we were writing a string class. We might start out like this:

These two variables have an intrinsic connection: m_length should always equal the length of the string held by m_string. If m_length were public, anybody could change the length of the string without changing m_string (or vice-versa). This would put the class into an inconsistent state, which could cause all sorts of bizarre problems. By making both m_length and m_string private, users are forced to use whatever public member functions are available to work with the class (and those member functions can ensure that m_length and m_string are always set appropriately).

We can also help protect the user from mistakes in using our class. Consider a class with a public array member variable:

If users can access the array directly, they could subscript the array with an invalid index, producing unexpected results:

However, if we make the array private, we can force the user to use a function that validates that the index is valid first:

In this way, we’ve protected the integrity of our program. As a side note, the at() function of std::array and std::vector do something very similar!

Benefit: encapsulated classes are easier to change

Consider this simple example:

While this program works fine, what would happen if we decided to rename m_value1, or change its type? We’d break not only this program, but likely most of the programs that use class Something as well!

Encapsulation gives us the ability to change how classes are implemented without breaking all of the programs that use them as well.

Here is the encapsulated version of this class that uses functions to access m_value1:

Now, let’s change the class’s implementation:

Note that because we did not alter the prototypes of any functions in our class’s public interface, our program that uses the class continues to work without any changes.

Similarly, if gnomes snuck into your house at night and replaced the internals of your TV remote with a different (but compatible) technology, you probably wouldn’t even notice!

Benefit: encapsulated classes are easier to debug

And finally, encapsulation helps you debug the program when something goes wrong. Often when a program does not work correctly, it is because one of our member variables has an incorrect value. If everyone is able to access the variable directly, tracking down which piece of code modified the variable can be difficult (it could be any of them, and you’ll need to breakpoint them all to figure out which). However, if everybody has to call the same public function to modify a value, then you can simply breakpoint that function and watch as each caller changes the value until you see where it goes wrong.

Access functions

Depending on the class, it can be appropriate (in the context of what the class does) for us to be able to directly get or set the value of a private member variable.

An access function is a short public function whose job is to retrieve or change the value of a private member variable. For example, in a String class, you might see something like this:

getLength() is an access function that simply returns the value of m_length.

Access functions typically come in two flavors: getters and setters. Getters are functions that return the value of a private member variable. Setters are functions that set the value of a private member variable.

Here’s an example class that has getters and setters for all of its members:

In this class, there’s no problem with allowing the user to directly get or set any of the member variables, so a full set of getters and setters is provided. In the MyString example above, no setter was provided for variable m_length because we don’t want the user to be able to set the length directly (length should only be set whenever the string is changed).

Rule: Only provide access functions when it makes sense for the user to be able to get or set a value directly.

Although you will sometimes see getter functions returning a non-const reference to a member variable, this should generally be avoided, as it violates encapsulation by allowing the caller to change the internal state of the class from outside of the class. It’s better if your getters return by value or const reference, and use setters to set state.

Rule: Getters should usually return by value or const reference, not non-const reference


As you can see, encapsulation provides a lot of benefits for just a little bit of extra effort. The primary benefit is that encapsulation allows us to use a class without having to know how it was implemented. This makes it a lot easier to use classes we’re not familiar with.

8.5 -- Constructors
8.3 -- Public vs private access specifiers

82 comments to 8.4 — Access functions and encapsulation

  • Nitin


    As a way of saying thank you for your invaluable site, I'll list out the typos I stumbled upon in some of your chapters, starting with this chapter.

    Typo - "In order to the use the class"  --> In order to use the class.

  • Sam

    Typo: "We’d break not only this program, but likely most of programs that use class Something as well!" -> "We’d break not only this program, but likely most of THE programs that use class Something as well!"

  • Royd

    Hi Alex. Really excellent material. I notice you (quite suddenly I think) refer to "users (of the object)" in the context of encapsulation. I'm not quibbling about the nomenclature, but nodding to how important it becomes when there are "users" of the class, i.e. other programmers, not, well, "users" I guess. It's a huge subject but it might be a chapter in its own right to summarise how these and other mechanisms avoid multiple programmers stepping on each others toes! You have quite rightly referred to this type of issue many times, but perhaps there is a case for drawing those issues together? A new student, unused to many people working on the same code may be surprised at the importance of this. (I think Matteo was getting at the same point).

  • Hi,

    I know it's a little off the beaten track but wouldn't it be better to create a class using the class wizard in Visual Studio and then accessing it through the header file created for that class?  Also, are virtual destructors necessary?

    • Hi Nigel!

      > better to create a class using the class wizard in Visual Studio
      Better, no. Faster, probably. But this tutorial isn't about IDE's, it's about the language. Most IDE's offer a way to generate code.

      > are virtual destructors necessary?
      Depends on your class. If the classes have memory that's not automatically managed, then yes, virtual destructors are necessary. Without them you'll get memory leaks.

  • pete

    Hi Alex,
    i've noticed the bodies of the access functions (mainly the setters and getters) are written with the same line of their declarations. is it something you would recommend to implement or just a means of demonstration?
    thanks again.

    • Alex

      I typically put access functions definitions on the same line as the declarations to help keep the code vertically compact. This works because the bodies of the functions are so trivial (typically a single line).

      I'm not aware of any reason to do otherwise, but your stylistic tastes may vary...

  • WiseFool

    I think it would sure be nice if, along with having a constructor and destructor with each class that -  especially in std library classes - it would become a standard programming practice to include a function named help(), that when invoked, would simply print a list of public functions and members in that class, with parameters and a brief description of how they worked or what they were for.  At least in utilty classes like cin, cout and string where it made sense to have a help() function.

    • nascardriver

      Hi WiseFool!

      That's what documentations are for

      • WiseFool

        Thank you, nascardriver, especially for the first reference.  It gives excellent coverage of what's in there and how it works, with even sample programs.  In just the few minutes that I looked at it this time I learned 2 or 3 new "tricks" that might be very handy to know some time.  (I was aware of those 2 sites and had used them for quick explanations or syntax, but hadn't stumbled on the detailed coverage they give of std library classes and functions.)

    • Donlod

      Well i started learning programming with java and eclipse as IDE. And the java api is very well documented, what the methods are doing, what the params are for what return value to expect etc. and this is all build into the ide on mouseover. When I started with c++ this was no longer the case since the std lib functions mostly only provide type of params and return but no further information and sometimes those are even hard to read or some fancy typedefs. For example i still do not get all those makros and typedefs in the windows api, in the end the typedefs are just some kind of int and char. I think this is what I am missing the most when looking at java (as for now).

      • nascardriver

        The standard library and Windows API are both documented, the documentations are just not integrated in the IDE. This would be a nice feature.

        > fancy typedefs
        You've probably encountered the STL (Standard template library).

        > i still do not get all those makros and typedefs in the windows api
        If you don't need Windows specific code don't use the Windows API. It's old code and inconsistent at times.

        * (Better examples than cppreference, but slower)

        • Donlod

          > If you don't need Windows specific code don't use the Windows API. It's old code and inconsistent at times.
          I was looking into xinput and windows to map xbox controller inputs to mouse and keyboard buttons.

  • Matteo

    I think that the best way to show the advantage of Encapsulation is by saying that it allows a clear separation between a user and a developer of a class. The developer is the only with the knowledge needed to work with the internals, and he/she offers an interface for the users to interact with such internals according to some use cases.

  • Luhan

    Can you implement a class from other file such as we do with the functions, for example,  the forward declaration in the header file and the definition in a .cpp file?

  • AMG

    Hey Alex,
    Typo: "Consider a class with an public array" -> "Consider a class with a public array". Your tutorial creates an illusion that programming is smooth, simple and easy.

  • Ivar

    In 8.2 — Classes and class members, you say that we should
    "Use the struct keyword for data-only structures. Use the class keyword for objects that have both data and functions."

    My question is, when would one have data-only structures? Doesn't this break the idea of encapsulation?

    • Alex

      That's a surprisingly hard question to answer. Generally, you'll use structs when you need a light-weight (non-encapsulated) way to move data as a unit, and classes otherwise. It does break the idea of encapsulation, but sometimes that's okay, especially if we are defining something that isn't reusable, to solve the specific case of moving data from one place to another. In such a case, defining a fully encapsulated class can be a bit overkill.

  • Hamed O.Khaled

    Hey Alex!!
    Can we say that the difference between the Abstraction and Encapsulation that:
    In Abstraction we know little things that help us to use whatever is through an interface i.e.  we know by pressing button on remote control specific functions will happen like changing channels or changing sound level.
    In Encapsulation [[Access restricted]] we don't know anything about internals even we have no controls over them ((changing sound level through button as was in the Abstraction))
    i.e. In T.V we don't know how is the display of the screen ((talking relative to the pixels)) is and we can't change in the mechanism on it.
    am I Right ?
    Thanks for this fabulous tutorials:)

Leave a Comment

Put all code inside code tags: [code]your code here[/code]