michael@0: # This Source Code Form is subject to the terms of the Mozilla Public michael@0: # License, v. 2.0. If a copy of the MPL was not distributed with this michael@0: # file, You can obtain one at http://mozilla.org/MPL/2.0/. michael@0: michael@0: =head1 NAME michael@0: michael@0: mpi-test - automated test program for MPI library michael@0: michael@0: =head1 SYNOPSIS michael@0: michael@0: mpi-test [quiet] michael@0: mpi-test list michael@0: mpi-test help michael@0: michael@0: =head1 DESCRIPTION michael@0: michael@0: The B program is a general unit test driver for the MPI michael@0: library. It is used to verify that the library works as it is michael@0: supposed to on your architecture. As with most such things, passing michael@0: all the tests in B does not guarantee the code is correct, michael@0: but if any of them fail, there are certainly problems. michael@0: michael@0: Each major function of the library can be tested individually. For a michael@0: list of the test suites understood by B, run it with the michael@0: I command line option: michael@0: michael@0: mpi-test list michael@0: michael@0: This will display a list of the available test suites and a brief michael@0: synopsis of what each one does. For a brief overview of this michael@0: document, run B I. michael@0: michael@0: B exits with a zero status if the selected test succeeds, or michael@0: a nonzero status if it fails. If a I which is not michael@0: understood by B is given, a diagnostic is printed to the michael@0: standard error, and the program exits with a result code of 2. If a michael@0: test fails, the result code will be 1, and a diagnostic is ordinarily michael@0: printed to the standard error. However, if the I option is michael@0: provided, these diagnostics will be suppressed. michael@0: michael@0: =head1 RESTRICTIONS michael@0: michael@0: Only a few canned test cases are provided. The solutions have been michael@0: verified using the GNU bc(1) program, so bugs there may cause problems michael@0: here; however, this is very unlikely, so if a test fails, it is almost michael@0: certainly my fault, not bc(1)'s. michael@0: michael@0: =head1 AUTHOR michael@0: michael@0: Michael J. Fromberger michael@0: Thayer School of Engineering, Hanover, New Hampshire, USA