Regression Test Instructions V0.1 --------------------------------- This collection of test suites is still being developed. Some of the subdirectories of the regression/ directory have not yet been merged into the script responsible for running the tests. Future versions of this collection will merge them and provide increased functionality for running the tests: All the tests that are currently merged into the collection with the exception of the AIM3 tests and the manufacturing memory tests can be run from a single script -- runall. The script takes an optional argument and a single parameter. The optional argument, -a, tells runall to run all suites in parallel instead of serially as it normally does. The parameter specifies the number of iterations to use for each test suite. Example: % runall -a 25 # run tests in suite in parallel 25 iterations The runall script will put the test output in a subdirectory of the regression/ directory called Report/ and name the test output after the name of the test suite subdirectory with the suffix .out added. In the runall script, there is a line beginning with SUBTESTS=. This specifies which test suites will be run by the script. Edit the line to run the particular subset of test suites you are interested in. Remember that if you run X tests, you will need to set your DISPLAY environment variable appropriately before invoking runall. After the tests have completed, you can generate a test report summary using the genreport script. It will look through the test output files in the Report/ directory and count the number of errors it finds in each one. The AIM3 test suite is currently working but not fully integrated. To run this test, change into the aim3/ directory and execute the multiuser program. You will be prompted with some test parameters and then the test will begin execution. The manufacturing memory test suite is currently working but not fully integrated. To run it, change into the mfg/ directory and invoke the mem program as follows: % mem -aimrwv -s <#>m Where <#> is the number of megabytes of memory you want allocated for the tests. The memory tests run forever, so when you are ready for them to stop running, type 'killall mem' to halt the tests. The next version of runall will incorporate the above two tests and will provide a more flexible method of running tests. It will take a run file as input which will specify which tests to run and how to run them. To get the latest version of the tests, grab the file homegrown:~billd/regression.tar.Z Send questions, comments, or suggestions to billd. - Bill Dorsey