Hi Darren, Performing performance unit tests with Cactus is one area I'd also like Cactus going into. However, I've not researched it yet. I remember this page talking about this too: http://www.kimble.co.uk/papers/grinder.htm
I haven't run your code yet (I plan to). Have you been able to trace why some calls take longer? Is it related to Cactus or to the app server? Does it do this on the different containers? The only issue I can see with Cactus would be if you run several tests in parallel (that would cause problems because test results would overwrite each other). But I don't think this is what you're doing. WRT Maven reports: the results are the standard JUnit results, right? So I guess the standard stylesheet would work. However I'm sure it can be improved. It would be quite easy to provide an additional property of the Maven Cactus plugin that allows to use the styledir attribute of the junitreport task. Thanks -Vincent > -----Original Message----- > From: Darren Hartford [mailto:[EMAIL PROTECTED] > Sent: 07 April 2004 16:44 > To: Cactus Users List > Subject: Junitperf on Cactus/Maven > > Hey all, > Sharing my experience and looking for some feedback. > > I have successfully used JunitPerf on top of the cactus-maven-plugin. > Unfortunately, the results give me some doubt (the timed measures seem too > sporadic) and I am hoping some of you more active testers could test and > give your feedback to share with the rest of us less experienced testers > on how to do performance/stress in-container testing. =-) > > > Here is some sample code to get people rolling: > > ===code=== > public class SimplePerfTestCase extends ServletTestCase { > public SimplePerfTestCase(String name) { > super(name); > } > > public static Test suite() { > TestSuite suite = new TestSuite(); > suite.addTest(simpleLoadTest(5, 10, "testNothing")); > return suite; > } > > public void testNothing(){ > assertTrue(true); > } > > public void setUp(){} > public void tearDown(){} > > private static Test simpleLoadTest(int users, int iterations, String > testMethod) { > //if average is more than 2 seconds (per user/iteration), fail > long maxElapsedTime = (2000 * users * iterations); > > Test testCase = new SimplePerfTestCase(testMethod); > Test loadTest = new LoadTest(testCase, users, iterations); > Test timedTest = new TimedTest(loadTest, maxElapsedTime); > > return timedTest; > } > } > ===code=== > > Unfortunately, I'm not sure how to extend the results to a report plugin > in Maven for the site generation task, but that's step two if this would > accurately measure performance. > -D > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
