We haven't put much thought into external developers using start_test for performance testing, so it's certainly possible that we have some baked-in assumptions about our testing infrastructure.
That code basically does a `git rev-parse HEAD` to get a SHA to associate with the current test session. If your test directory isn't in in a git repo, that code will break, though that particular error message should be innocuous. As a lame/temp workaround, you can try something like https://github.com/chapel-lang/chapel/compare/master...ronawho:fix-perfsha-no-git , but if you are trying to do performance testing outside the chapel-lang git repo, we can try to improve start_test's usability. Elliot ?-----Original Message----- From: "Stavenger, Timothy J" <[email protected]> Date: Wednesday, March 28, 2018 at 2:31 PM To: "[email protected]" <[email protected]> Subject: start_test --performance failing on computePerfStats Resent-From: Brad Chamberlain <[email protected]> Resent-To: Elliot Ronaghan <[email protected]> Resent-Date: Wednesday, March 28, 2018 at 2:30 PM I¹m using the Chapel language¹s start_test to run unit & performance tests on my project¹s code (as we anxiously wait for the official Chapel test framework). The unit tests appear to be working fine. The performance tests run, but fail at https://github.com/chapel-lang/chapel/blob/release/1.16/util/start_test#L936-L943 <https://github.com/chapel-lang/chapel/blob/release/1.16/util/start_test#L936-L943>. I¹m running this in Jenkins in a chapel/chapel:1.16.0 Docker container. In my Jenkinsfile I export the ŒCHPL_TEST_PERF_DIR¹ env variable, Œcd¹ into the test folder, and then run: start_test --performance The thought was I¹d start about as simple as possible, but I get the error: [Error: Failed to save current sha to /var/lib/jenkins/jenkins-home/workspace/HPDA/AHM/chgl/test/performance/dat/perfSha.dat] The file exists in Jenkins¹ workspace at least after the build finishes. It looks like the start_test script is trying to execute the ŒcomputePerfStats¹ util, but is failing without logging what the problem is. Granted this is rather vague, but do you have an idea of what may be going wrong? Or at least some direction as to where to look? Is there a way to turn on verbose logging for start_test to get a better idea of what is happening? Thanks! ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Chapel-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/chapel-users
