The implementation presented was written and compiled using the GNU C++ compiler, g++, version 2.7.0. (At the time of writing, the authors believe that the WARPED system will compile under g++ version 2.6.3, but not under 2.5.8 or earlier.) It should also be noted that some of the classes that are defined in libg++ are used within the applications that are distributed with the WARPED system, so if these applications are to be used, libg++ will be required as well.
WARPED was designed to run on top of the MPI message protocol. (The developers used the mpich implementation of the MPI standard, version 1.0.10. Information on mpich is available from http://www.mcs.anl.gov/home/lusk/mpich/index.html ) It is possible to develop sequential applications with WARPED without having MPI installed on a system, but in order to use WARPED in parallel, MPI must be installed. It is also worth noting that the functions used from MPI are very simple point to point communication, and that most message passing libraries would have similar functions. Thus, porting WARPED to a different message passing library should be a relatively simple endeavor.
The WARPED system is intended for two types of users - those wishing to build distributed applications model as discrete event simulations, and those wishing to experiment with TimeWarp itself. The application level user needs no knowledge of MPI or internal details of the simulation kernel, but the TimeWarp experimenter most likely will.