Hello, I've a suggestion on my own, since MPI allows people to develop massively parallel applications, libraries itself should also be taking advantage of this. Today, we are at a point where we are talking about big data, but in fact, we are about to face with huge data (I think this will be a fact in a couple of years). So, people like me already working on implementing parallel solutions for daily problems.Still, the data itself is not the only one who becomes huge, but the source code too. With the newest discoveries of algorithms, patterns, technologies in development, companies and individuals are doing their best to take advantage of it and this causes to produce huge source code piles which take hours to compile. So, my suggestion is that, as standard, MPI compilers (like mpicc, mpif90) should offer an option to the users which enables them to start a distributed compile operation, it can be executed either on multiple cores locally or multiple CPUs throughout the network. With the best wishes, -- Özgür Pekçağlıyan Computer Engineer (M.Sc.)