I downloaded FiPy months ago and this week I decided to test its parallel feature. So I tried to run the mesh1D.py example I got from github using
$ mpirun -np 2 python examples/diffusion/mesh1D.py
Two viewers pop up, each solving half of the problem like it was supposed to be. After the first problem is completed (explicit transient diffusion), it asks me to press return to proceed.
True
True
Explicit transient diffusion. Press <'return> to proceed...Explicit transient diffusion. Press <'return> to proceed...
But nothing happens when I press return. I know little about parallel programming, but is it normal that both processors are asking to press return? Shouldn't this part of the code be handled by only one core?
Some extra information: the mesh1D example works fine if I don't use mpi. And I have run the parallel.py test as well, which works correctly.
I'm using python 2.7.14, Trilinos 12.12.1, FiPy 3.2 and mpi4py 3.0.0
This is a bug. Thanks for reporting it.