Tokenized substring search

[Generic Usenet Account]

I had a need to tokenize a string, and this is what I came up with. Is
this the most efficient way of tokenizing, or is a more efficient
solution possible?

Thanks,
Gary

#include <iostream>
#include <string>
using namespace std;
main()
{
string::size_ty pe pos;
string cmd="a1::b1::c1 ::d1";
string subStr;
string token="::";
cout << "cmd string = " << cmd << endl;
try
{
for(pos=0; pos != string::npos;)
{
pos = cmd.rfind(token );
if (pos != string::npos)
{
cout << "pos = " << pos << '\n';
cmd.erase(pos);
cout << "substring cmd = " << cmd << endl;
}
else
{
throw pos;
}
} // end for
} //end try
catch(string::s ize_type noPos)
{
cerr << "no more tokens in cmd" << endl;
}
}

[Evan]

There are two pre-built solutions I know of that you can use instead if
you want to avoid writing it yourself. The C standard library (and thus
C++ library) comes with strtok which will probably work but has its own
problems. Also, there is a free library (I think under a BSD-ish
licence) called Boost at boost.org that has a tokenizer that is built
using C++ features and fixes these problems.