Thu, 22 Jan 2015 13:21:57 +0100
Incorporate requested changes from Mozilla in review:
https://bugzilla.mozilla.org/show_bug.cgi?id=1123480#c6
michael@0 | 1 | # This Source Code Form is subject to the terms of the Mozilla Public |
michael@0 | 2 | # License, v. 2.0. If a copy of the MPL was not distributed with this |
michael@0 | 3 | # file, You can obtain one at http://mozilla.org/MPL/2.0/. |
michael@0 | 4 | |
michael@0 | 5 | =head1 NAME |
michael@0 | 6 | |
michael@0 | 7 | mpi-test - automated test program for MPI library |
michael@0 | 8 | |
michael@0 | 9 | =head1 SYNOPSIS |
michael@0 | 10 | |
michael@0 | 11 | mpi-test <suite-name> [quiet] |
michael@0 | 12 | mpi-test list |
michael@0 | 13 | mpi-test help |
michael@0 | 14 | |
michael@0 | 15 | =head1 DESCRIPTION |
michael@0 | 16 | |
michael@0 | 17 | The B<mpi-test> program is a general unit test driver for the MPI |
michael@0 | 18 | library. It is used to verify that the library works as it is |
michael@0 | 19 | supposed to on your architecture. As with most such things, passing |
michael@0 | 20 | all the tests in B<mpi-test> does not guarantee the code is correct, |
michael@0 | 21 | but if any of them fail, there are certainly problems. |
michael@0 | 22 | |
michael@0 | 23 | Each major function of the library can be tested individually. For a |
michael@0 | 24 | list of the test suites understood by B<mpi-test>, run it with the |
michael@0 | 25 | I<list> command line option: |
michael@0 | 26 | |
michael@0 | 27 | mpi-test list |
michael@0 | 28 | |
michael@0 | 29 | This will display a list of the available test suites and a brief |
michael@0 | 30 | synopsis of what each one does. For a brief overview of this |
michael@0 | 31 | document, run B<mpi-test> I<help>. |
michael@0 | 32 | |
michael@0 | 33 | B<mpi-test> exits with a zero status if the selected test succeeds, or |
michael@0 | 34 | a nonzero status if it fails. If a I<suite-name> which is not |
michael@0 | 35 | understood by B<mpi-test> is given, a diagnostic is printed to the |
michael@0 | 36 | standard error, and the program exits with a result code of 2. If a |
michael@0 | 37 | test fails, the result code will be 1, and a diagnostic is ordinarily |
michael@0 | 38 | printed to the standard error. However, if the I<quiet> option is |
michael@0 | 39 | provided, these diagnostics will be suppressed. |
michael@0 | 40 | |
michael@0 | 41 | =head1 RESTRICTIONS |
michael@0 | 42 | |
michael@0 | 43 | Only a few canned test cases are provided. The solutions have been |
michael@0 | 44 | verified using the GNU bc(1) program, so bugs there may cause problems |
michael@0 | 45 | here; however, this is very unlikely, so if a test fails, it is almost |
michael@0 | 46 | certainly my fault, not bc(1)'s. |
michael@0 | 47 | |
michael@0 | 48 | =head1 AUTHOR |
michael@0 | 49 | |
michael@0 | 50 | Michael J. Fromberger <sting@linguist.dartmouth.edu> |
michael@0 | 51 | Thayer School of Engineering, Hanover, New Hampshire, USA |