Is this a bug of boost::tokenizer ?

Here is the code. When

Code:

string line("dd-9c32d00@3");

is changed to

Code:

string line("dd-9c32d00");

The program crashes.
It seems that tok.end() does not bound the end of string parsing?

Code:

#include <boost/tokenizer.hpp>                                                               
using namespace std;                                                                                                            
using namespace boost;

int main(){
    //string line("dd-9c32d00@3");
    string line("dd-9c32d00");// crashes!
    typedef boost::tokenizer< boost::char_separator<char> > token;
    boost::char_separator<char> sep("@");
    token tok(line, sep);
                                                                                                                                           
    token::iterator itt=tok.begin();                               
    if (itt==tok.end())                                                                                                        
        cout<<*itt; 
    else                                                                                                         
        cout<<*(++itt);
                                                                                                                                                                             
    return 0;                   
}

My question actually is, how to tell if 'itt' is the last piece delimited by the token? I compared it with tok.end(). If false it is not, otherwise it is.

Read the example.

If the iterator is pointing to the end of the tokenizer, then you *don't* want to access the iterator, as it does not contain a valid token.