int vs unsigned int

[benbehr]

With the following code I get a warning that I am comparing signed and unsigned int.

vector<float> test[10];

for (int i=0; i<test.size(); i++){...} 

On the topic I have found this (http://www.cplusplus.com/forum/beginner/8529/) which I think is quite ok, however the last two replies have confused me a bit.

In the case of loops, it is better to make sure that i is of the same type as SIZE. In the case of std containers use the appropriate size_type defined by the container rather than int or unsigned int.

How do I find the type in question?
What is it for a vector?

Thanks,
Ben

[Catfish]

http://www.cplusplus.com/reference/stl/vector/
Scroll down to Member Types to see size_type.

It is used as such, in your example:
for (std::vector<float>::size_type i=0; i < test.size(); ++i);

You may omit adding the std:: if you have using namespace std;.

If you use C++11 (the newest standard of C++, late 2011), you can simplify this loop greatly:
for (float &f: test);
You compiler might not support the so-called range-based for() loop, though.

[benbehr]

Thanks Catfish,
the first version works fine for me. But why do I have specify what type the vector members have? Shouldn't the size always be the same type?

Regarding the simplified loop I guess you are right and my compiler is not up to it. In any case apart from that I also don't quite understand it. I would assume it gets the type from the used vector "test", but that should also mean that it knows that the vector contains floats, right?
Would I have to use 'f' in your example (e.g. in test[f])?
What is the '&' in this case?

Thanks,
Ben

[Catfish]

But why do I have specify what type the vector members have?

Because std::vector is a template, and it needs that information to generate the actual code.

Shouldn't the size always be the same type?

I think it probably is, in practice. I could compile that loop with std::vector<bool>::size_type instead, without an error being issued.

As for the range-based for loop... the f takes the value of each element in the vector.
The & means that f is a reference: this means that when you change f inside the loop, it will change in the vector as well.

for (float &f: test)
    f = 33.0f; // update all values to 33.0f

for (float f: test)
    std::cout << f << std::endl; // print all values from test 

[benbehr]

Ok, so let me see if I got this right, but first one new question. In f = 33.0f, what does the second 'f' stand for, and why can it be written to a float?

Now to the understanding part.
This means in fact if my verctor holds a struct I could do the following?

#include "std_lib_facilities.h"

struct testStruct{
   // Definition of struct members
   float a;
   int b;
   string c;
};

int main(){
vector<testStruct> test;
testStruct temp;

// adding some more values to the vector
for (int i=0; i<3; i++){
    temp.a = i*2.4;
    temp.b = i;
    temp.c = "someting";
    test.push_back(temp);
}

for (testStruct &f: test){
   f.a = f.a+3;
   f.b = f.b+2;
   f.c = f.c+" and "+f.c // i know that is kind of not so ....
}

}

So, if I assume that the above is right (as I cant test it). How would I implement something where I try to access the previous value? Something like:

if(i>0) test[i].a = test[i-1]+3;

Thanks!

[Catfish]

In f = 33.0f, what does the second 'f' stand for, and why can it be written to a float?

33 is a simple value, and is by default treated as an int.
Therefore I added .0f to make sure it will be treated as a float.
I personally consider it to be good practice but it's not always needed to be done.
So the f doesn't get written anywhere, it's just convention.

Read more here: http://www.cplusplus.com/doc/tutorial/constants/

So, if I assume that the above is right (as I cant test it). How would I implement something where I try to access the previous value?

It compiles just fine (after editing for the includes of course, and adding the missing semicolon).
The previous value? No, in that case a range-based for loop is not what you should use. You're better off using a regular for loop.

[benbehr]

Thank a lot for the help! I think this has answered all my questions.
Tusend Tak (danish for 1000 thanks - something they say here)

[Catfish]

Good luck with C++11, if you go that way.

[benbehr]

Thanks, but for now I will stick with what I have. I have a fast approaching deadline on the project. But maybe for future.
So from your point of view, is it a blessing or rather not?

[Catfish]

Blessing, mostly. It has many usability improvements, like the special for loop, automatic type deduction, and initializer lists, which would allow you to initialize your float vector like this:
vector<float> test = {0.4f, 33.0f, 144.54f};

In spite of this, I'd rather divert you to the D language.