Pointers and links [duplicate]

Links, like pointers in C ++, are addresses of objects in memory. Being able to refer to a specific object is so important that some languages ​​(Java, for example, or the .NET family) state that this is practically the only way to work with objects. At this level of abstraction, there is no difference between references and indexes. The difference appears a level lower: a pointer is a value, an object of the first class, roughly speaking, a memory cell number. Some arithmetic operations can be made with them, compared as numbers, etc. In order to have a pointer (i.e., an address) to access an object at that address, a dereference operation is applied:

 *p = 100; // указатель p разыменовывается, и в // полученную ячейку памяти записывается число 100 

References, in turn, do not possess the semantics of "object addresses", but the semantics of the objects themselves. You may think that the link is a pointer that itself, automatically, applies a dereference. From this there are several consequences: for example, the immutability of references. In C ++, there is simply no link assignment syntax; any such assignment would be an assignment to the object it points to.

Now for your examples: *p=&t; incorrectly (an attempt to assign a pointer to a number to a dereferenced pointer p (specifically to the number t), not the number itself), but *p=t absolutely correct (refer to the address stored in the index and write the number t) .

If we talk about deep thoughts, then, first of all, the pointers came from C and remained a heavy legacy. I think incorrect memory addressing is the most common cause of C ++ programs crashing. Links cannot replace pointers (they are not objects, unlike pointers), but they make life easier when you need to pass some structure weighing a couple of kilobytes into a function, and copying will slow down the program, but you don’t want to mess around with pointers. Links have the same usage syntax as the objects to which they refer, and therefore are more or less interchangeable.

I feel that I have already written a lot, but I could not formulate anything clear :(
I hope it became a little clearer

PS In C ++, there is also the concept of "smart pointers" that behave like pointers, but there are fewer problems with them. When you understand the links and the usual pointers, I highly recommend them to look.

1. The link and the pointer are in fact no different, except that the link is initialized once and it cannot be redefined + the syntax is different. For example, if a reference to a class, then a point is used to refer to members. instead of ->
2. How is * p = t not correct? Quite to yourself is correct. If p points to some int variable, then nothing prevents it from doing this (write the value from t to the variable to which p points)
3. What difficulties do you have in mind? Why pointers?
4. I understand that the difficulty is that you do not know why you need pointers? Well, such an example:

 struct hello { int value1; int value2; }; struct hello instance; void myfunction(struct hello * p) { printf("value1 is %d\n", p->value1); p->value2 = 777; } int main(void) { instance.value1 = 1; instance.value2 = 2; myfunction(&instance); return 0; } 

As you can see, we pass a pointer to the structure to the function. Do not copy the data itself through the stack. In addition, if there were no pointer, then we would not be able to change the value value2.

Speaking of complexity, I’m talking about the existence of pointers and links. why are they even needed? why it is impossible to manage simply variables?

When I studied C ++, I myself had the exact same question. I then just continued to read the book until I encountered the use of pointers in practice. I re-read this chapter, once again tried to penetrate, and it turned out to understand. Then, when I started working with graphics, studied OpenGL and DirectX, I met with these pointers everywhere, and there were even pointers to pointers. I then endured the brain. Then I figured it out and got used to it. Now pointers seem to me something familiar and necessary.

If there were no libraries that use pointers, I think we could do without them. But this would suffer performance, ease of development and clarity of the program.

Pointers or links are used:

1. To transmit an address of a complex structure or object to avoid copying. For this purpose, constant links are best suited for syntax.

2. To return multiple values ​​from a function. A single return value, through return, from a function may not be enough. There are two ways to return several values: use a structure with several fields (which is very inconvenient if these values ​​are not related by meaning), or writing to the address passed through the pointer. In the second method, you can use the link, but syntactically it looks like passing a parameter, not a return. Therefore, the pointer in this case better shows the programmer's intentions in the code. The use of global variables I will not even call the third way. This is a bad tune of programming (why, you will know yourself when you gain experience), and it is better to never do this.

3. Links are needed for copy constructors. You will learn about them when you get to classes.

4. All variables are placed on the stack. They have their own scope, when they exit from which they are destroyed. To control the life of a variable, you may need to allocate it from dynamic memory with new and save a pointer to this memory area so that it can be accessed and deleted.

5. Pointers are used in data structures such as lists. You will learn about lists after studying classes. They should be in any normal C ++ book.

In general, the applications of mass. I do not think that I listed at least a tenth of them.

The link from the pointer differs as follows:

1. The link syntax is simpler (p instead of * p or p.value instead of p-> value)
2. A link cannot be “zero”, that is, it must be initialized with something.
3. The link cannot be redefined, that is, it will always point to the same memory location.

See also the article about links in C ++ on wikipedia.

In your second question, everything is really correct, as cy6erGn0m said :)

Why do we need links at all - an excellent question, in my opinion :)

In general, of course, the code with links can always be rewritten into code with pointers. Links were entered into C ++ (they were not in C) precisely to simplify the syntax.

Another advantage of links is that it is easier to modify the code. For example, you have a function that accepts an object of some type A:

 void f(A a); 

At some point, it becomes clear that objects of type A are too large, and copying them every time you call f () is too expensive. In order not to overwrite the function and its places of call to pointers, you can simply change the declaration of the function:

 void f(A& a); 

or better yet:

 void f(const A& a);