Question

I’m trying to avoid reimplementing my own clumsy versions of standard algorithms, and thus am playing with the standard library versions. Since I’m no expert in C++, I proceed with caution and full debug options switched on.

Specifically, I’m using a binary search on a valarray container. The following code block seems to produce correct results, and valgrind does not complain. Still, I do feel I’m on a slippery slope as I’m not sure if what I’m doing is really allowed or am I just being let out by the compiler.

A representative piece of code:

#include <iostream>
#include <valarray>
#include <algorithm>
#include <typeinfo>

using namespace std;

int main(){

 valarray<double> v(10);
 for (int i=0 ; i<10 ; ++i){
   v[i]=2. *i ; 
   cout<<v[i]<<"  ";
 }
 cout << "\n";

 double what=17;
 double* it=lower_bound(&v[0], &v[10],what) ; 

 cout<<it-&v[0]<<" "<<typeid(&v[0]).name()<<" ";
 cout<<typeid(it).name()<<" "<<typeid(it-&v[0]).name()<<"\n"; // ???

 int idx=it-&v[0];
 cout<<"v["<<idx<<"]="<<v[idx]<<"\n";
}

Questions:

1. Is what I’m doing here really legal?
2. How come the difference between two pointers to double becomes an int? (in the line with the ??? comment)
3. What is the overhead for the type conversion? — I am concerned with efficiency as this sort of functionality is going to sit in the part of the code which takes more than 90% of the computation time.

Answer 1

1. You’re using an int to index into the valarray. That’s valid for the example, but not in general. Use an std::size_t to index into a valarray. (The same goes for std::vector and ordinary arrays.)

2. The difference between two pointers to any type is of an unspecified integer type, likely int or long and always small enough to fit in an std::ptrdiff_t.

3. Which conversion?