Compare with size_t, return int?

Question

I'm writing some code examples from "How to Think Like a Computer Scientist in C++", and this one is about handling playing-card type objects and decks. I'm facing this situation:

int Card::find(const std::vector<Card>& deck) const {
    size_t deckSize = deck.size();
    for (size_t i=0; i<deckSize; i++)
        if (equals(*this, deck[i])) return i;

    return -1;
}

I couldn't use ".length()" on a vector in C++ in Visual Studio 2010 as in the text, and instead had to use .size() which returns (I believe) std::size_type. I figured I could use size_t and get away with it in order to avoid problems on different architectures, as I've been reading, but I'm wondering if I return i, but it's bigger than an integer, will I crash the program?

[Edited to be more specific in my question:] Once I start using vectors for larger things than cards, I considered using unsigned int because of a compiler mismatch warning, but I feel returning an unsigned int or int has a few issues: 1) int will not take a sufficiently large vector index. 2) returning unsigned int will not let me return -1. 3) unsigned int isn't equal to size_t on all architectures (I'm also doing microcontroller programming on an ARM Cortex-M3).

What should I do if I ever have a large enough vector?

You should consider getting a different C++ book. There is a list of [good introductory C++ books](http://stackoverflow.com/questions/388242/the-definitive-c++-book-guide-and-list). I'm not familiar with _How to Think Like a Computer Scientist in C++_, but its first example has a `void main()` and uses ``. It doesn't look promising. — James McNellis, Dec 16 '10 at 04:38

score 3 · Accepted Answer · answered Dec 16 '10 at 04:34

3

Casting from size_t to int will not "crash" your program, but it's a bad bad practice. On the other hand, STL includes nice find algorithm for what you are doing.

answered Dec 16 '10 at 04:34

Nikolai Fetissov

82,306
11
110
171

I'm leery about using the STL for non-PC programming since I'm also learning to be able to write for microcontrollers (specifically the ARM Cortex-M3). It seems the STL has the capability to balloon code size if handled improperly. – darvelo Dec 16 '10 at 18:45
1

You might want to stick to C for these sort of things then, though many here might start yelling :) – Nikolai Fetissov Dec 16 '10 at 18:48

score 2 · Answer 2 · answered Dec 16 '10 at 04:37

2

int is 32 bit on 32 / 64 bit Windows and Linux. i will get truncated if greater than two at the 31st. you could use unsigned int and your program will be fine unless storing more than 4 G elements in the vector :)

answered Dec 16 '10 at 04:37

Cristian

21
1

Thanks for the suggestion. It's just that if I want to return -1 in case no results return, unsigned int would not handle it. I'll amend my post to reflect that. – darvelo Dec 16 '10 at 18:54

score 0 · Answer 3 · answered Dec 16 '10 at 04:33

0

size_t is typically an unsigned int but you can't rely on that. If it's bigger than an int you won't crash, you'll just overflow into a (probably negative) number.

Assuming you're not going to have several tens of thousands of cards in one vector, I'd be happy returning the int.

answered Dec 16 '10 at 04:33

Robert

6,412
3
24
26

2

size_t will probably not be an unsigned int on 64-bit platforms. – Aaron Klotz Dec 16 '10 at 04:43
Exactly. That's why you can't rely on that. – Robert Dec 16 '10 at 05:46

score 0 · Answer 4 · answered Dec 16 '10 at 05:22

0

You can also return std::pair<size_t, bool>, similar to std::map insert(). Second template argument means success or fail.

If you ok with this, you could also use boost::optional

answered Dec 16 '10 at 05:22

dimba

26,717
34
141
196

Compare with size_t, return int?

4 Answers4