>>>>> "E" == Emil Ong <onge / mcs.anl.gov> writes:

E> I've been working on this MPI Ruby thing and in it, I embed ruby into an
E> MPI program.  To do this, I run ruby_run().  I was having a lot of
E> problems because every MPI program needs to run MPI_Finalize() before
E> exiting and I was calling this after ruby_run(), but since
E> ruby_run() calls exit(), the MPI_Finalize() wasn't being reached.  I put
E> it into an atexit(), but why does ruby_run() call exit anyway?

 The good question is why do you want to call ruby_run() ?

 The goal of ruby_run() is precisely to run only once and call exit().


Guy Decoux