Skip to content

miravassor/CPP_Pool

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

C++ Pool - 42

This repository is a comprehensive collection of Object-Oriented Programming (OOP) exercises in C++, designed to span from foundational to advanced programming concepts. As you navigate through the modules, you'll engage with various aspects such as container usage, class and inheritance implementation, polymorphism, iterators, and algorithm development.

Below are my notes, filled with essential information, and the learnings I've gathered on my journey of exploring the vast and powerful paradigm of OOP. These notes are crafted with the hope of aiding newcomers in navigating the complexities of OOP with greater ease and understanding.

Useful Links

An introduction to C++ and OOP

  • Different than procedural programming → can lead to spaghetti code, harder to maintain and expand, lot of function parameters
  • Combines related variables and functions into an object
    • variables = properties
    • functions = methods

4 Pillars

  1. Encapsulation
  2. Abstraction
  3. Inheritance
  4. Polymorphism

I - Encapsulation

  • Less and sometimes no parameters to functions → easy to expand or delete features, allowed by parameters that are direct properties of the object
    • “ The best functions are those with no parameters”

II - Abstraction

  • No need to understand everything to use objects → we can hide some properties and methods in an object
  • Leads to object with simpler interface
  • Reduce the impact of change

III - Inheritance

  • Define a base object with common properties and methods and let other objects inherit this attributes
  • Eliminate redundant code

IV - Polymorphism

  • Allow for inherited objects to define or add unique behavior to the base object

  • Eliminate if/else forest and redundant anesthetic switch cases with simple code :

    object.execute() // will work for every object

Implementations

  • Separation of Interface and Implementation:
    • Keeps the class definition (header file) clean and focused on the interface, making it easier for other programmers to understand the class's API without getting bogged down in implementation details.
  • Reduced Compilation Dependency:
    • When a function is defined outside the class definition, any change to the function's implementation does not require recompiling all the source files that include the class's header file.
    • This can significantly reduce build times in large projects.

Encapsulation

Encapsulation is a fundamental concept in object-oriented programming (OOP). It involves bundling the data (variables) and the methods (functions) that operate on the data into a single unit, which in most OOP languages, is the class. Here are key aspects of encapsulation:

Data Hiding: Encapsulation allows restricting direct access to some of an object's components, which is a means of preventing accidental interference and misuse of the methods and data. This is usually achieved using access specifiers like private, protected, and public in C++.

Controlled Access: Through encapsulation, an object can control its internal state and can change it through specific methods known as getters and setters. This controlled access safeguards against unauthorized access and modification of data.

Modularity: By encapsulating related properties and behaviors within classes, a program's code becomes modular, making it easier to understand, maintain, and debug.

Simplifying Complexity: By hiding the internal workings of objects and exposing only what is necessary, encapsulation simplifies the complexity of software development. Users of a class don't need to understand its internal complexities to use it.

Abstraction: Encapsulation aids in abstraction by allowing the creation of a more abstract representation of a model. It emphasizes the behavior of an object, not how the behavior is implemented.

This approach leads to code that is more secure and robust, as well as easier to understand and maintain.

  • this is a pointer that refers to the current instance (object) of a class. It is used within class member functions to access the members (variables and functions) of the current instance

Canonical form

class Span {
	private:
		;
	public:
		Span();
		Span(unsigned int N);
		Span(const Span &tocopy);
		~Span();

		Span& operator=(const Span &rhs);
};

Declaration

Span::Span() : _N(1) {}

Span::Span(unsigned int N) : _N(N) {}

// Span::Span(const Span &tocopy) : _N(tocopy._N), _v(tocopy._v) {}
// equals to (for auto-memory management or not heap objects):
// but less efficient: additionnal steps!!
Span::Span(const Span &tocopy) {
	*this = tocopy;
}

Span::~Span() {}

Span& Span::operator=(const Span &rhs) {
	if (this != &rhs) {
		this->_N = rhs._N;
		this->_v = rhs._v;
	}
	return (*this);
}
  1. Default Constructor

    Assignation syntax

    Animal::Animal(void) : _type("Animal")
  2. Destructor: To deal with resource de-allocation.

  3. Copy Constructor: To handle copying of resources correctly.

  4. Copy Assignment Operator: To handle assignment of resources correctly.

    Animal &Animal::operator=(Animal const &copy) {
    	if (this != &copy)
    		*this = copy;
    	return (*this);
    }

Operator surcharge

std::ostream&   operator<<(std::ostream& *out*, const Class& obj);
bool operator>(const Fixed &rhs) const;
bool operator<(const Fixed &rhs) const;
bool operator>=(const Fixed &rhs) const;
bool operator<=(const Fixed &rhs) const;
bool operator==(const Fixed &rhs) const;
bool operator!=(const Fixed &rhs) const;
Fixed operator+(Fixed const &rhs) const;
Fixed operator-(Fixed const &rhs) const;
Fixed operator*(Fixed const &rhs) const;
Fixed operator/(Fixed const &rhs) const;
// prefix increment
T& operator++(int); // int = convention
T& operator--(int);
// postfix increment
T operator++();
T operator--();

// definition example
bool Fixed::operator>(const Fixed &rhs) const {
	return ((this->_fixPointValue > rhs.getRawBits()) ? true : false);
}

Classes (example)

#pragma once

#include ...

class Bureaucrat {
	private:
		const std::string	_name;
	public:
		Bureaucrat(void);
		Bureaucrat(const std::string name, unsigned int grade);
		Bureaucrat(Bureaucrat const &copy);
		~Bureaucrat(void);
		Bureaucrat&			operator=(Bureaucrat const &rhs);
		const std::string	getName(void) const;
		void				upGrade(void);

		class	GradeTooHighException {
			public:
				virtual void what() const throw() {
					std::cerr << "Grade too high" << std::endl;
				}
		};
};

std::ostream&	operator<<(std::ostream& out, const Bureaucrat& bureaucrat);

Pointer vs Reference

both pointers and references are used to work with memory addresses and access objects indirectly

  • Pointers → *
    • need to use the address-of operator (&) to obtain the address of an object that you want to point to
int* ptr; // Declaration of a pointer
int x = 10;
ptr = &x; // Assign the address of 'x' to the pointer
  • References → &
    • don't need to explicitly obtain an address because a reference is an alias to an existing object
int y = 20;
int& ref = y; // Declaration and initialization of a reference

Pointers vs. References in Function Parameters

Pointers are typically used for functions that need to modify the object they point to or work with arrays.

References are often used for function parameters when you want to pass an object by reference to avoid making a copy or when you want to enforce that the parameter cannot be null.

void modifyValue(int* ptr) {
    (*ptr)++; // Modify the value pointed to by 'ptr'
}

void modifyValue(int& ref) {
    ref++; // Modify the value referred to by 'ref'
}

Examples

std::string	str = "HI THIS IS BRAIN";
std::string	*stringPTR = &str;
std::string	&stringREF = str;

std::cout << &str << std::endl;
std::cout << stringPTR << std::endl;
std::cout << &stringREF << std::endl;
std::cout << str << std::endl;
std::cout << *stringPTR << std::endl;
std::cout << stringREF << std::endl;
0x7fffcccf1618
0x7fffcccf1618
0x7fffcccf1618
HI THIS IS BRAIN
HI THIS IS BRAIN
HI THIS IS BRAIN

Returning *this (reference convention)

why do i need to return the *this and not this ?

Returning *this instead of this in the assignment operator is an important convention in C++ that relates to the types involved and how they are used in expressions.

  • this is a pointer to the current object.
  • *this is a reference to the current object itself.

The reason you return *this (a reference) rather than this (a pointer) is to enable the natural and expected usage of assignment in C++ expressions. When you write code like a = b = c;, you're relying on the assignment operator to return a reference to the left-hand object (a and b in this case), not a pointer. This allows the assignment to chain naturally, as each assignment operation returns a reference to the object it just assigned to, which then becomes the left-hand operand of the next assignment.

For example, consider the following code:

Span a, b, c;
a = b = c;

Here's how it works:

  1. b = c calls b.operator=(c). This method modifies b to be a copy of c and returns a reference to (this).
  2. a = (b = c) is now effectively a = b (because b = c returned a reference to b). So, a.operator=(b) is called, which modifies a to be a copy of .

If the assignment operator returned a pointer (this), the code a = b = c; would not work as expected, because you can't assign a pointer to a Span object directly to another object. The types wouldn't match up, and the natural chaining of assignments would be broken.

So, by returning *this, you're adhering to C++ conventions and enabling your class to be used in a way that C++ programmers would expect.


Null-ability

Pointers can be null or uninitialized, which means they may not point to a valid object. You must check for null pointers before de-referencing them to avoid undefined behavior.

References must always be initialized when declared and cannot be null. They always refer to an existing object.


Reassignment

Pointers can be reassigned to point to different objects or null pointers.

References cannot be reassigned after initialization. They remain bound to the object they were initialized with.

Syntax for Accessing Values:

Pointers use the de-reference operator (*) to access the value they point to.

References are used directly, without any additional syntax, to access the value they refer to.

int value = *ptr; // Accessing the value through a pointer
int value = ref; // Accessing the value through a reference

In-class accessibility

  • private: Accessible only within the class it's defined. Not child!
  • public: Accessible from anywhere the object is visible.
  • protected: Middle ground, adds the ability for child classes to access.
  • ** Protected Access Level**:
    • Within the Class: Same as private, can use directly.
    • In Derived Classes: Can access protected members of the base class.
    • Outside the Class: Can't access, similar to private. Useful for inheritance. Allows child classes to use and modify these members, which are hidden from the rest of the program

const

  1. const Member Functions (at end):
    • Syntax: ReturnType FunctionName() const;
    • These functions do not modify any member variables of the class (except those marked as mutable).
    • Usage: Allows the function to be called on const instances of the class. Ensures the function won't change the object's state.
  2. const Member Variables:
    • Syntax: const DataType VariableName;
    • These variables must be initialized when the object is created (at construction) and cannot be changed afterward.
    • Usage: Useful for defining immutable properties of the class.
  3. const in Constructor Parameters:
    • Syntax: ClassName(const DataType& ParameterName);
    • Used to prevent modification of parameters passed by reference or pointer.
    • Usage: Protects input parameters from being accidentally modified inside the constructor.
  4. const Return Types:
    • Syntax: const ReturnType FunctionName();
    • Prevents the caller from modifying the returned value, useful when returning references or pointers.
    • Usage: Ensures the integrity of the returned data, especially when returning references to internal class members.

static

  • Static Member Variables:

    • Syntax: static DataType VariableName;
    • Belong to the class, not any particular object instance.
    • Shared by all instances of the class.
    • Initialized once and retain their value between function calls.
    • Usage: Useful for values that should be shared across all instances, like counters or global settings within the class context.
  • Static Member Functions:

    • Syntax: static ReturnType FunctionName(Parameters);
    • Also belong to the class rather than any object instance.
    • Can only access static members and other static functions, not non-static members.
    • Usage: Often used as utility functions that relate to the class but don't need to interact with instance-specific data.

    When you define a static member function outside the class in C++, you don't use the static keyword in the function definition. The static keyword is only used in the class declaration to specify that the member function is static. Here's why:

    1. Class Declaration: Inside the class declaration, the static keyword is necessary to tell the compiler that the function is a static member of the class. This affects how the function is called (i.e., it can be called without an instance of the class) and how it behaves (i.e., it does not have access to instance-specific data).
    2. Function Definition: When you define the function outside the class, you don't need to repeat the static keyword. At this point, the compiler already knows the function is static from the declaration in the class.
  • Static Local Variables in Functions:

    • Syntax: static DataType VariableName; within a function.
    • Declared within a function.
    • Retain their value between function calls (preserve state).
    • Initialized only once, the first time the function is called.
    • Usage: Commonly used for implementing functions that need to remember a state or value between calls, such as in the case of a function that counts how many times it has been called.

mutable

mutable int mutableVar;

  • Normally, all member variables of a const object are read-only; they cannot be changed once the object is created.
  • However, if a member variable is declared with the mutable keyword, it can be modified even if it belongs to a const object.

virtual

  • Indicates that a member function can be overridden in a derived class.
  • Essential for achieving polymorphism in C++.
  • When you declare a function as virtual in a base class, you are telling the C++ compiler that you want to allow derived classes to provide their own specific implementation of that function.

virtual destructor:

Here's why a virtual destructor is important in a class hierarchy:

  • When you delete a derived class object through a base class pointer without a virtual destructor in the base class, the destructor of the derived class is not called. This can lead to resource leaks if the derived class is managing resources (like dynamic memory).
  • A virtual destructor ensures that the correct destructor is called for derived class objects, even when they are referred to by a base class pointer or reference.

virtual ~Animal(void); ensures that when an object of a derived class (that inherits from Animal) is destroyed through a pointer or reference to Animal, both the destructor of the derived class and that of Animal are called in the correct order, thereby releasing all resources correctly.


explicit

  • Applied to constructors to prevent automatic type conversion.

Pure virtual function

virtual void show() = 0; // Pure virtual function

  1. Abstract Class: A class with at least one pure virtual function becomes an abstract class. You cannot instantiate an abstract class directly.
  2. Interface Requirement: The pure virtual function acts as a contract. Derived classes must provide their own implementation of this function. If they don't, they too become abstract classes.
  3. Polymorphism: This is used to achieve polymorphism. A base class pointer or reference can be used to refer to objects of derived classes. When you call a pure virtual function through a base class pointer or reference, the version defined in the derived class is invoked.

Type casting

Implicit cast: all the data types of the variables are upgraded to the data type of the variable with largest data type. → can lose information, signs (signed/unsigned), overflow (long long → float)

Cast operators

  • static_cast : compile-time cast, allows casting from any pointer type to void pointer and vice versa, can call Class operators, not secure (does not verify validity of the cast)

    → Converting between related types, such as between base and derived classes or between numeric types

    • static_cast<*dest_type*>(source);
  • **dynamic_cast** : mainly used for safe down-casting at run time, must be one virtual function in the base class or “Source type is not polymorphic” compiler error

    safest for downcasting in class hierarchies and throws a bad_cast exception (or returns a null pointer in case of pointers) if the cast is not valid.

    • dynamic_cast<new_type>(source)

    Downcasting: Casting a base class pointer (or reference) to a derived class pointer (or reference) is known as downcasting

    Upcasting: Casting a derived class pointer (or reference) to a base class pointer (or reference) is known as upcasting.

  • **const_cast** : un-const an object

    • const_cast<int>(source);
    const int num = 10;
    int* modifiable = const_cast<int*>(&num);
    // *modifiable = 20; // Undefined behavior if you modify the original const value
  • **reinterpret_cast** : convert any pointer type to any other pointer type, even if the types are unrelated, treats the sequence of bits (the object representation) of one thing as if it were a sequence of bits of another type, without checking the safety or validity of such a conversion. Compile time.

    low-level reinterpreting of bit patterns


try / throw / catch

  • try: This block tests a set of statements for errors.
  • throw: When an error is detected, the throw statement is used to send an exception.
  • catch: This block catches and handles exceptions thrown by the try block.

Inside class definition of what the what.e() function will call:

class	AnException : public std::exception {
			public:
				virtual const char* what() const throw() {
					return "Error [...]";
				}
		};

Inside a function:

if (condition) {
		throw Class::AnException();

Handling the exception:

try {
        // do stuff
}
catch (const std::exception &e) {
	// if stuff fails
    std::cerr << e.what() << std::endl;
}

Components

virtual const char* what() const throw()

virtual: indicates that the method can be overridden in a derived class. In the context of an exception class, virtual ensures that if a derived class provides its own implementation of what(), that implementation will be used instead of the base class version when the method is called through a reference or pointer to the base class

const char*: return type of the method. const char* means the function returns a pointer to a constant character array (C-style string). The content of this string typically describes the error that caused the exception. This return type is chosen for compatibility reasons, as C++ exceptions need to be usable even in contexts where C++ standard library types like std::string might not be available or desirable

what(): name of the method. what() is a standard method in C++ exceptions, used to provide a human-readable description of the exception.

**const :** method does not modify state of the object

throw(): exception specification. throw() indicates that the function is guaranteed not to throw any exceptions. In later versions of C++, this notation is deprecated in favor of noexcept, which serves a similar purpose. For the what() method, indicating that it will not throw any exceptions is important because this method is often called in a catch block, and throwing another exception in such a block would be problematic.

  • Using the library example

    #include <stdexcept>
    #include <string>
    
    class MyException : public std::runtime_error {
    public:
        MyException(const std::string& message)
            : std::runtime_error(message) {
        }
    };

Templates

Allows to define generic classes or functions

Warning : one typename by type!

Function templates

Write a function that can operate on different data types

template <typename T>
void swap(T& a, T& b) {
    T temp = a;
    a = b;
    b = temp;
}

Class templates

Define generic classes. For example, a Box class that can store any type of item.

template <typename T>
class Box {
private:
    T content;
public:
    void setContent(T newContent) {
        content = newContent;
    }
    T getContent() {
        return content;
    }
};

Template Specialization

define a specific implementation for a particular data type. For example, a specialized version of Box for int

template <>
class Box<int> {
// Special implementation for int
};

Multiple typename (for multiple types and no cast when calling)

// apply function pointer to each index of an array
template <typename T, typename U, typename V>
void	iter(T *add, U size, V fp) {
	for (size_t	i = 0; i < size; i++)
		fp(add[i]);
}

Containers

The Containers library is a generic collection of class templates and algorithms that allow programmers to easily implement common data structures like queues, lists and stacks

There are two (until C++11), now three (since C++11) classes of containers:

  • sequence containers
  • associative containers
  • unordered associative containers (C++11)

The container manages the storage space that is allocated for its elements and provides member functions to access them, either directly or through iterators (objects with properties similar to pointers).

Most containers have at least several member functions in common, and share functionalities. Which container is the best for the particular application depends not only on the offered functionality, but also on its efficiency for different workloads.

Sequence Containers

Sequence containers implement data structures which can be accessed sequentially.

Container Description
array (cppreference.com) static contiguous array (class template)
vector (cppreference.com) dynamic contiguous array (class template)
deque (cppreference.com) double-ended queue (class template)
forward_list (cppreference.com) singly-linked list (class template)
list (cppreference.com) doubly-linked list (class template)

Associative Containers

Associative containers implement sorted data structures that can be quickly searched (O(log n) complexity).

Container Description
set (cppreference.com) collection of unique keys, sorted by keys (class template)
map (cppreference.com) collection of key-value pairs, sorted by keys, keys are unique (class template)
multiset (cppreference.com) collection of keys, sorted by keys (class template)
multimap (cppreference.com) collection of key-value pairs, sorted by keys (class template)

Iterators

standard containers provide various types of iterators to enable different ways of traversing the container

iterator: This is a bidirectional iterator for a container, allowing both forward and backward traversal

const_iterator: Similar to iterator but provides read-only access to the container's elements

reverse_iterator: This is an iterator that moves backward through the container (from the end to the beginning).

const_reverse_iterator

5 types of iterators (defined in Standard Template Library)

  • Input Iterators:

    • These iterators can read from the pointed-to element.
    • They are the least powerful but most flexible iterators.
    • They support operations like increment (moving to the next element) and dereferencing (accessing the element).
    • Commonly used in single-pass algorithms and for reading from streams.
  • Output Iterators:

    • These are used for writing to the pointed-to element.
    • Like input iterators, they support increment and dereferencing, but for writing instead of reading.
    • Commonly used for single-pass write algorithms and writing to streams.
  • Forward Iterators:

    • They have all the capabilities of input iterators.
    • They can read and write (if they are not constant iterators) and move forward (like input and output iterators), but they can also go through the sequence multiple times (unlike input and output iterators).
    • Suitable for multi-pass algorithms on a sequence of elements.
  • Bidirectional Iterators:

    • These iterators extend forward iterators with the ability to move backward (decrement).
    • They can be used in algorithms that need to iterate forwards and backwards through a sequence.
    • Common in containers like std::list, std::set, std::map, etc.
  • Random Access Iterators:

    • The most powerful iterators, which combine the capabilities of bidirectional iterators with the ability to directly access any element in a sequence (like an array).
    • They support a wide range of operations, including addition, subtraction, and comparison of iterators.
    • Used with containers like std::vector and std::deque.
  • Examples

    1. Input Iterator

    Used for reading elements in a forward direction. Commonly used with input streams.

    #include <iostream>#include <iterator>#include <vector>int main() {
        std::vector<int> vec = {1, 2, 3, 4, 5};
        std::istream_iterator<int> start(std::cin), end; // Input iterator from standard input
        std::copy(start, end, std::back_inserter(vec)); // Reads until end-of-file or invalid input
    
        for (int n : vec) {
            std::cout << n << " ";
        }
    }

    2. Output Iterator

    Used for writing elements. Common with output streams.

    #include <iostream>#include <iterator>#include <vector>int main() {
        std::vector<int> vec = {1, 2, 3, 4, 5};
        std::ostream_iterator<int> out_it(std::cout, ", "); // Output iterator to standard output
        std::copy(vec.begin(), vec.end(), out_it); // Writes elements of vec to standard output
    }

    3. Forward Iterator

    Can read/write and move forward, suitable for multiple passes.

    #include <forward_list>#include <iostream>int main() {
        std::forward_list<int> flist = {1, 2, 3, 4, 5};
        for (std::forward_list<int>::iterator it = flist.begin(); it != flist.end(); ++it) {
            std::cout << *it << " "; // Reading using forward iterator
        }
    }

    4. Bidirectional Iterator

    Can move forwards and backwards. Common in std::list, std::set, etc.

    #include <iostream>#include <list>int main() {
        std::list<int> mylist = {1, 2, 3, 4, 5};
        for (std::list<int>::reverse_iterator rit = mylist.rbegin(); rit != mylist.rend(); ++rit) {
            std::cout << *rit << " "; // Reverse iteration using bidirectional iterator
        }
    }

    5. Random Access Iterator

    Supports direct element access, addition, subtraction, and comparison.

    #include <iostream>#include <vector>int main() {
        std::vector<int> vec = {1, 2, 3, 4, 5};
        std::vector<int>::iterator it = vec.begin() + 2; // Direct access
        std::cout << "Third element: " << *it << std::endl;
    }

Files interface

#include <iostream>
#include <fstream>

std::ifstream	in;
std::ofstream	out;

Function pointers

// typedef
typedef void (Harl::*t_func) (void);

// point to the functions
t_func	fptr[] = {&Harl::debug, &Harl::info};

// call the function
this->*fptr[i])()

Switch case

  1. switch Expression: The switch statement evaluates an expression, typically a variable.
  2. case Labels: These are specific values that the expression is compared against. If the expression matches a case value, the code following that case label is executed.
  3. break Keyword: Typically, a break statement is used at the end of each case block. It exits the switch statement, preventing the execution from falling through to the next case.
  4. default Label: This is an optional label that executes if none of the case labels match. It's like an "else" in an "if-else" ladder.
void	Harl::complain(std::string level) {
	std::string	levels[] = {"DEBUG", "INFO", "WARNING", "ERROR"};
	int	i = 0;

	while (i < 4 && levels[i].compare(level))
		i++;
	switch (i) {
		case 0:
			this->debug();
		case 1:
			this->info();
		case 2:
			this->warning();
		case 3:
			this->error();
			break;
		default:
			std::cout << "A very human thing to make problems out of anything." << std::endl;
	}
}
  • Omitting break can be useful in certain situations where you want multiple case labels to execute the same code block, but this should be used carefully to avoid errors.
  • The case labels must be constant expressions, typically literals or const variables.

Ternary

condition ? expression1 : expression2;


Heap allocation

  • new operator is used for dynamic memory allocation on the heap
  • delete operator to free the heap when new was used
  • delete [] free multiples allocations

Random

#include <cstdlib>
#include <ctime>

srand(time(NULL)); // initialise random number generator
int i = rand(); // store pseudo-random number
// rand() % 100 would generate a random number between 0 and 99

(srand(time(NULL)) → NULL is used as an argument to the time function. The time function, when passed NULL, returns the current calendar time (the number of seconds since the Unix epoch, January 1, 1970). The NULL in this case indicates that you don't need to store the result in a time structure, but are only interested in the numeric value it returns.


Libs & co

  • std::string string class, offers a dynamic and flexible way to handle strings without worrying about memory management
  • std::flush explicitly flush or clear the output buffer, the output buffer is a temporary storage area where data is collected before being sent to the actual output device, such as the console or a file. When you write data to an output stream using << the data is typically not immediately sent to the output device but buffered for performance reasons. This buffering can improve efficiency by reducing the number of write operations to the output device. However, there are times when you may want to ensure that the data in the buffer is immediately written to the output device, and that's where std::flush comes in
  • std::cin.good() is used to check the state of the input stream std::cin to determine if the input stream is in a good (valid) state for further input operations

About

Learning C++ the 42 way

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published