Community
Participate
Working Groups
When running the headless "emma" command in a Jenkins job, I often get this error [java] INFO: Starting test session MyTest [java] INFO: Running test... ... [java] INFO: ...OK [0.201s] [java] INFO: Tests finished. [java] INFO: Elapsed time: 1.556 seconds. [java] INFO: Total number of tests executed: 2 [java] INFO: Successful tests: 2 [java] INFO: Failed tests: 0 [java] INFO: Errors: 0 [java] INFO: Ignored tests: 0 [java] INFO: Overall status: OK [java] ERROR: No coverage has been collected during coverage launch (code 5101). the strange thing is that most of the time the coverage.exec is actually created (in spite of the ERROR above)... and indeed, the build does NOT issue a failure. by looking at buckminster code, Emma.java, I see that in case there is no coverage session, buckminster throws an exception with message "No coverage sessions found", while the above error is issued by eclemma itself, and the faq (http://www.eclemma.org/faq.html#trouble01) says "This happens when the Java process did not terminate properly, e.g. has been manually killed with the Terminate button. Make sure your program terminates on its own. " indeed, the buckminster's emma command terminates, in fact, by using DEBUG I see INFO: Tests finished. INFO: Elapsed time: 69.02 seconds. INFO: Total number of tests executed: 113 INFO: Successful tests: 113 INFO: Failed tests: 0 INFO: Errors: 0 INFO: Ignored tests: 0 INFO: Overall status: OK ERROR: No coverage has been collected during coverage launch (code 5101). Doing full workspace refresh Waiting for jobs to end thus, I assume the buckminster command terminates correctly, but internally EclEmma issues that error because its Java process is terminated before it has finished? If I run plain junit test suite and the plug-in junit test suite, the first suite reports this error If I run plug-in junit test suite first and then plain junit test suite, none of them reports this error when the error is reported, the coverage.exec file is generated anyway (most of time, when it's not generated, buckminster itself fails, but I rarely experienced this, say 1 out of 50 builds). Can it be due to some synchronization issues when buckminster runs eclemma and does not wait for it to terminate? The above problem happens in a Jenkins server, which is almost 10 years old, thus not very powerful...
