I use openmpi and linux mint, consider the following example:
#include <boost/mpi.hpp>
#include <iostream>
#include <string>
#include <boost/serialization/string.hpp>
namespace mpi = boost::mpi;
using namespace std;
int main()
{
mpi::environment env;
mpi::communicator world;
if (world.rank() == 0)
{
world.send(1, 0, std::string("3664010"));
while (1)
{
world.send(1, 0, std::string("3664012"));
sleep(1);
}
}
else
{
std::string msg;
string dst;
bool first = true;
while (1)
{
world.recv(0, 0, msg);
if (first) {dst = msg;first = false;}
std::cout << "slave received=" << dst << " msg=" << msg << std::endl;
}
}
return 0;
}
Compilation: mpic++ -std=c++0x test.cc -lboost_serialization -lboost_mpi
Run: mpirun -np 2 ./a.out
Output: slave received=3664010 msg=3664010
slave received=3664012 msg=3664012
slave received=3664012 msg=3664012
Bug is only reproduced when all message have equal length. If second message for example will be «3664012andmore» everything work fine:
slave received=3664010 msg=3664010
slave received=3664010 msg=3664012andmore
slave received=3664010 msg=3664012andmore
slave received=3664010 msg=3664012andmore
It looks like dst and msg use the same memory buffer. And they start to use different memory buffer only if strings length are different. I use following workaround(msg = string()) to tell compiler that msg is changed:
std::cout << «slave received=» << dst << " msg=" << msg << std::endl;
msg = string();
And it works fine. Есть некостыльные решения? Спасибо.