Consider the following example:
#include <iostream>
#include <string>
#include <vector>
#include <iterator>
int main()
{
std::string sen = "abc def ghi jkl";
std::istringstream iss(sen);
std::vector<std::string>
vec(std::istream_iterator<std::string>(iss),
std::istream_iterator<std::string>());
std::copy(vec.begin(), vec.end(),
std::ostream_iterator<std::string>(std::cout, "\n"));
}
The compiler throws an error when called std::copy
request for member 'begin' in 'vec', which is of non-class type...
I can get around the error, for example:
std::istream_iterator<std::string> it_begin(iss);
std::istream_iterator<std::string> it_end;
std::vector<std::string> vec(it_begin, it_end);
or by placing parentheses around each parameter, for example:
std::vector<std::string>
vec((std::istream_iterator<std::string>(iss)),
(std::istream_iterator<std::string>()));
or even with the new uniform initialization in C ++ 11:
std::vector<std::string> vec { , };
Why does the compiler parse the declaration in the example as a function declaration? I know about the most unpleasant analysis, but I thought this only happened with empty parameter lists. I also wonder why the second workaround works.
source
share