diff options
Diffstat (limited to 'c++.html.markdown')
-rw-r--r-- | c++.html.markdown | 63 |
1 files changed, 59 insertions, 4 deletions
diff --git a/c++.html.markdown b/c++.html.markdown index 80ad3a6c..176ea1a8 100644 --- a/c++.html.markdown +++ b/c++.html.markdown @@ -72,7 +72,7 @@ void func(); // function which may accept any number of arguments int* ip = nullptr; // C standard headers are available in C++. -// C headers end in .h, while +// C headers end in .h, while // C++ headers are prefixed with "c" and have no ".h" suffix. // The C++ standard version: @@ -553,10 +553,14 @@ Point Point::operator+(const Point& rhs) const return Point(x + rhs.x, y + rhs.y); } +// It's good practice to return a reference to the leftmost variable of +// an assignment. `(a += b) == c` will work this way. Point& Point::operator+=(const Point& rhs) { x += rhs.x; y += rhs.y; + + // `this` is a pointer to the object, on which a method is called. return *this; } @@ -757,7 +761,7 @@ failure: // things are a little cleaner, but still sub-optimal. void doSomethingWithAFile(const char* filename) { - FILE* fh = fopen(filename, "r"); // Open the file in read mode + FILE* fh = fopen(filename, "r"); // Open the file in shared_ptrread mode if (fh == nullptr) throw std::runtime_error("Could not open the file."); @@ -811,6 +815,57 @@ void doSomethingWithAFile(const std::string& filename) ///////////////////// +// Smart Pointer +///////////////////// + +// Generally a smart pointer is a class, which wraps a "raw pointer" (usage of "new" +// respectively malloc/calloc in C). The goal is to be able to +// manage the lifetime of the object being point to without explicitly deleting +// the object. The term itself simply describes a set of pointers with the +// mentioned abstraction. +// Basically smart pointers should preferred over raw pointers, to prevent +// risky memory leaks, which happens if you forget to delete the object. + +// Usage of a raw pointer: +Dog* ptr = new Dog(); +ptr->bark(); +delete ptr; + +// With the usage of smart pointers you dont have to worry about the deletion +// of a object anymore. +// A smart pointer describes a policy, to count the references on the +// pointer. As matter of fact the objects gets destroyed when the last +// reference on the object gets destroyed. + +// Usage of "std::shared_ptr": +void foo() +{ +// Its not longer necessary to delete the Dog. +std::shared_ptr<Dog> doggo(new Dog()); +doggo->bark(); +} + +// Beware of possible circular references!!! +// There will be always a reference, so it will be never destroyed! +std::shared_ptr<Dog> doggo_one (new Dog()); +std::shared_ptr<Dog> doggo_two (new Dog()); +doggo_one = doggo_two; // p1 references p2 +doggo_two = doggo_one; // p2 references p1 + +// As mentioned before there is a set of smart pointers. The way you have to +// use it, is always the same. +// This leads us to question, when to use which one? +// std::unique_ptr - use it when you just want to hold one reference on +// the same object. +// std::shared_ptr - use it when you want to hold multiple references on the +// same object and want to make sure that it´s de-allocated +// when all refences are gone. +// std::weak_ptr - use it when you want to hold multiple references from +// different places for references for which it´s no problem +// tp de-allocate. + + +///////////////////// // Containers ///////////////////// @@ -873,7 +928,7 @@ cout << ST.size(); // will print the size of set ST // Output: 0 // NOTE: for duplicate elements we can use multiset -// NOTE: For hash sets, use unordered_set. They are more efficient but +// NOTE: For hash sets, use unordered_set. They are more efficient but // do not preserve order. unordered_set is available since C++11 // Map @@ -902,7 +957,7 @@ cout << it->second; // Output: 26 -// NOTE: For hash maps, use unordered_map. They are more efficient but do +// NOTE: For hash maps, use unordered_map. They are more efficient but do // not preserve order. unordered_map is available since C++11. // Containers with object keys of non-primitive values (custom classes) require |