How to initialize an n-sized array after declaration in C++

Question

I'm making a text-based RPG kind of thing in C++, and I'm making the Inventory class. Here is my code:

class Inventory {
    public:
        int size;
        Item items[];

        Inventory(int size): size(size) {

        }
};

In the constructor, I need to set items to an array of Items with the length of size. I know in Java, I'd just do this.items = new Item[size], but I've been looking for a while now, and even on the cplusplus.com tutorial, it doesn't say how I'd do this. Is there a way? If so, how?

okay thanks, ill try that. One problem, however, is that I need to limit the size. Can I do this with vectors or would I have to manually limit it (e.g not letting it add more items to the vector once it's too big) — Zac Garby, Mar 05 '16 at 16:25
Depends on how you're actually adding `Items` to the interned vector/array. — πάντα ῥεῖ, Mar 05 '16 at 16:28

score 2 · Accepted Answer · answered Mar 05 '16 at 16:26

2

C++ doesn't support this regarding raw arrays. The intended thing in c++ is using a std::vector<Item> instead:

class Inventory {
    public:
        int size;
        std::vector<Item> items;

        Inventory(int size): size(size), items(size) {

        }
};

answered Mar 05 '16 at 16:26

πάντα ῥεῖ

1
13
116
190

why should we go for vector? we can solve this by simple 'new' and 'delete' right? – Gilson PJ Mar 05 '16 at 16:33
1

@GilsonPJ Getting `new`/`delete` right is complicated and peppered with pitfalls you never thought of. The c++ standard container classes were designed to overcome all of these for you. There's no need to reinvent the wheel all the time. – πάντα ῥεῖ Mar 05 '16 at 16:35
Dynamic allocation seems a bit over blown for that – ALOToverflow Mar 05 '16 at 16:35
@πάντα ῥεῖ If I were to use this, would I have to initialize `items`? If so, how – Zac Garby Mar 05 '16 at 16:50
@ZacGarby _"would I have to initialize items? "_ It's already done in my sample code. Are you asking about initializing individual items? – πάντα ῥεῖ Mar 05 '16 at 16:53
no, i just meant like usually you have to use `new`... i think. You do in Java anyway – Zac Garby Mar 05 '16 at 16:57
1

@ZacGarby: Java's `new` does a lot more than a C++ `new`. It is very unfortunate that Java's inventors thought it was a good idea to copy the keyword from C++. It has caused endless confusion among Java programmers who learn C++. – Christian Hackl Mar 05 '16 at 17:04
@ZacGarby: P.S.: Why would you use an array in Java and not an `ArrayList`? – Christian Hackl Mar 05 '16 at 17:04
In Java, I'd use an array if I wanted it to not be resizable, shorter to type out or easier to access values (arr[n] instead of arrList.get(n)) – Zac Garby Mar 05 '16 at 17:06
@ZacGarby Do you know the array size at compile time? Rather use a `std::array` then. – πάντα ῥεῖ Mar 05 '16 at 17:08

score 2 · Answer 2 · answered Mar 05 '16 at 16:52

Using std::vector<Item> is the one and only sane solution here.

In a comment to another answer, you are adding this piece of information:

One problem, however, is that I need to limit the size. Can I do this with vectors or would I have to manually limit it (e.g not letting it add more items to the vector once it's too big)

A std::vector by itself is pretty limitless. Its only real limit is the available memory.

But you should get into the habit of encapsulating container classes anyway. Usually, a container class offers way more functionality than you will ever need for one particular use case. For example, your items member variable may never require rbegin(), cend(), shrink_to_fit(), difference_type, get_allocator() or pop_back(), just to name a few examples.

It is therefore good practice to create a custom data type which offers only those operations which you really need, and implement the custom data type in terms of std::vector. It then becomes trivial to implement additional constraints.

Example:

#include <vector>
#include <string>
#include <stdexcept>
#include <exception>
#include <iostream>

// just a super-simple example:
struct Item {
    std::string name;
};

// wraps std::vector<Item>:
class Items {
public:
    Items(int max_size) :
        max_size(max_size),
        items()
    {
    }

    void Add(Item const& item) {
        if (static_cast<int>(items.size()) == max_size) {
            throw std::runtime_error("too many items");
        }
        items.push_back(item);
    }

    Item Get(int index) const {
        return items[index];
    }

private:
    int max_size;
    std::vector<Item> items;
};

int main()
{
    Items items(5);
    try {
        items.Add({ "sword" });
        items.Add({ "shield" });
        items.Add({ "torch" });
        items.Add({ "axe" });
        items.Add({ "boots" });
        std::cout << items.Get(3).name << "\n";
        items.Add({ "gloves" });
    } catch (std::exception const& exc) {
        std::cerr << exc.what() << "\n";
    }
}

Notice that this example uses exceptions to handle the error. This may not be suitable to your use case; you may consider assert instead.

Well, you took the burden at least to explain it to the end. Excellent (I was just too lazy now). Greetings from Bavaria my dear neighbour. — πάντα ῥεῖ, Mar 05 '16 at 16:57
@πάνταῥεῖ: It must be my old teaching-assistant spirit :) Thanks! — Christian Hackl, Mar 05 '16 at 17:00

How to initialize an n-sized array after declaration in C++

2 Answers2