Buffer size question.

[Swerve]

Hi,

I am trying to find the size of the buffer for an uninitialized string and an empty string, but using size() and length() doesn't seem to be helping me.

Here is the program I am using to try and find the size of the buffers being used for the strings:

#include <iostream>
#include <string>
using namespace std;
int main()
{
	string s1;  //unitialized string using size()
	cout << s1.size() << endl;

	string s2 = ("");  //empty string using size()
	cout << s2.size() << endl;

	string s3;  //unitialized string using length()
	cout << s3.length() << endl;

	string s4 = ("");  //empty string using length()
	cout << s4.length() << endl;

	system ("PAUSE");
	return 0;
}

Which is giving the following output:

0
0
0
0
sh: PAUSE: not found
Press Enter to continue!

Could anyone tell me how I can find the size of the buffers being used?

I'm unsure if I'm missing the point somewhat.

Many thanks!

[grumpy]

size() and length() return the length of the buffer that is actually being used. Since a std::string records the actual length of the content, there is no need for terminating characters in the buffer. That is why you will be seeing a string string initialised with "" ending up with a size() of zero.

If you want the length of the internally allocated buffer, use capacity(). Change capacity() manually using reserve().

[iMalc]

Could anyone tell me how I can find the size of the buffers being used?

I'm unsure if I'm missing the point somewhat.

What makes you think that in those examples the buffers are necessarily greater than zero anyway?

You're not allowed to write into a string past its length, so finding the buffer size is of dubious use anyway

[EVOEx]

You're not allowed to write into a string past its length, so finding the buffer size is of dubious use anyway

The only reason I can think of for finding the buffer size is optimization. If you know you're going to append dozens of strings it might be a lot better first to reserve enough data so that no appending actually has to allocate data. Otherwise it might have to re-allocate a lot of times.
So if I know up front the length of the resulting string, I tend to reserve the data before actually doing anything with it.

[anon]

The comments that you have an uninitialized string vs empty string are also wrong. The default constructor initializes the string object to represent an empty string.

[Swerve]

Thanks for all the information chaps. Learning a lot.

I have a question.

Using the following code:

#include <iostream>
int main()
{
	std::string s1;
	std::cout << s1.capacity() << std::endl;

	system ("PAUSE");
	return 0;
}

In Ubuntu/KDevelop the output is 0.

In Windows/V.Studio the output is 15.

Can anyone suggest why it happens?

I'm somewhat stumped by it's occurrence.

Thanks very much, much appreciated.

[anon]

It is entirely implementation-dependent. The capacity must be at least as much as the size of the string: it could be more, and often is (because the string also probably needs to meet some complexity requirements, when it grows), but doesn't have to be at any time.

If VC++ implementation uses a short-string optimization (short strings are stored on the stack), I suppose it is impossible for it to have zero capacity.

(Also include <string> for string class, and <cstdlib> for the system function.)