use resize instead of substr and reassigning

[IamRezaMousavi]

this avoids unnecessary copying

[UniQP]

The change looks fine but looking at the code I would expect that we resize the string to a size of SF_MAX_SEM_NAME_LEN rather than SF_MAX_SEM_NAME_LEN - 1.

[IamRezaMousavi]

This line does the same as the previous one.

[snicolet]

            -  shm_name = shm_name.substr(0, SF_MAX_SEM_NAME_LEN - 1);
            +  shm_name.resize(SF_MAX_SEM_NAME_LEN - 1);

Why did we use SF_MAX_SEM_NAME_LEN - 1 in master for the length instead of SF_MAX_SEM_NAME_LEN, by the way?

Edit: comments crossed with UniQP

[IamRezaMousavi]

            -  shm_name = shm_name.substr(0, SF_MAX_SEM_NAME_LEN - 1);
            +  shm_name.resize(SF_MAX_SEM_NAME_LEN - 1);

Why did we use SF_MAX_SEM_NAME_LEN - 1 in master for the length instead of SF_MAX_SEM_NAME_LEN, by the way?

Edit: comments crossed with UniQP

I'm not sure, but if you'd like I can modify it to use resize(SF_MAX_SEM_NAME_LEN).

[UniQP]

            -  shm_name = shm_name.substr(0, SF_MAX_SEM_NAME_LEN - 1);
            +  shm_name.resize(SF_MAX_SEM_NAME_LEN - 1);

Why did we use SF_MAX_SEM_NAME_LEN - 1 in master for the length instead of SF_MAX_SEM_NAME_LEN, by the way?
Edit: comments crossed with UniQP

I'm not sure, but if you'd like I can modify it to use resize(SF_MAX_SEM_NAME_LEN).

Please change it to resize(SF_MAX_SEM_NAME_LEN).

[Disservin]

i believe it was because we use c_str for the shm function calls and c_str is null terminated, so the actual amount of characters we can use is - 1

[UniQP]

i believe it was because we use c_str for the shm function calls and c_str is null terminated, so the actual amount of characters we can use is - 1

But then the check is wrong and needs to be changed to shm_name.size() >= SF_MAX_SEM_NAME_LEN.

[Disservin]

i believe it was because we use c_str for the shm function calls and c_str is null terminated, so the actual amount of characters we can use is - 1

But then the check is wrong and needs to be changed to shm_name.size() >= SF_MAX_SEM_NAME_LEN.

mhh that is true..

[UniQP]

Wait, if we use #define SF_MAX_SEM_NAME_LEN NAME_MAX and NAME_MAX is defined as the length without the null terminator, we don't need the -1. (I haven't checked the other #defines.)

[IamRezaMousavi]

the size() method returns the number of characters in the string, not including any null-termination.
so the check is ok and must be shm_name.size() > SF_MAX_SEM_NAME_LEN

here the example code:

#include <iostream>
#include <string>
#include <cstring>

#define MAX_LEN 5

int main()
{
    std::string s = "123456789";
    std::cout << "size is " << s.size() << '\n';
    
    if (s.size() > MAX_LEN) {
        s.resize(MAX_LEN);
    }
    
    std::cout << "now string is " << s << '\n';
    std::cout << "and c_str returns " << s.c_str() << '\n';
    std::cout << "and strlen returns " << strlen(s.c_str()) << '\n';

    return 0;
}

Its return:

size is 9
now string is 12345
and c_str returns 12345
and strlen returns 5 // <- we need to char[6] to store this string

[IamRezaMousavi]

What should I do?