Am 09.01.2015 um 20:51 schrieb Phil Steitz: > On 1/9/15 10:56 AM, sebb wrote: >> On 9 January 2015 at 17:21, Phil Steitz <phil.ste...@gmail.com> wrote: >>> On 1/9/15 10:06 AM, James Carman wrote: >>>> On Fri, Jan 9, 2015 at 12:03 PM, Phil Steitz <phil.ste...@gmail.com> wrote: >>>> >>>>> No, must be some issue with the environment or JDK. I get >>>>> consistent, stable success under both Ant and Maven using the 1.5 >>>>> JDK above. >>>> Our CI builds are all fine, right? >>> I can't explain the results linked above. I have no idea what maven >>> does under the covers to try to simulate a 1.5 build in that >>> environment or what 1.5 jdk is actually being used. The tests >>> succeed consistently for me using Oracle 1.5-1.7 JDKs under both Ant >>> and Maven. I did have to down-rev Maven to 3.1.1, get rid of some >>> cruft and tweak the Ant build to grab the Hamcrest jars; but the >>> test goals succeed consistently. >> Maven does not simulate anything. >> Instead of running compile and test under its own JDK, it uses the >> alternate jdk. >> >> http://maven.apache.org/plugins/maven-compiler-plugin/examples/compile-using-different-jdk.html >> >> I've made this all clearer in the commons-parent-pom docs: >> >> http://commons.apache.org/commons-parent-pom.html >> and >> http://commons.apache.org/building.html >> >> [May not yet have been published] >> >> Note that Jenkins only failed the tests *once*; other tests (using the >> same JDK) have been OK. >> As I wrote elsethread, that suggests the tests are not 100% stable. >> Or perhaps there is an issue on a specific Jenkins host; I've not >> checked which hosts were involved yet. > > I just looked at a few of the failed tests. I suspect that what is > going on is that tests from the same class are being executed > concurrently and the test class has instance data. I did not think > the pom was set up to allow this and I have never seen this kind of > failure before. > > Phil
When building the source distribution using the Java 1.5 profile on Windows 8.1 I get the test failures below. ISTR that I had similar results for previous releases (the OOME is a different story). According to my Maven settings, a 1.5.0_21 Java SDK should be used. With Java 1.7 there are no test failures. Oliver Results : Failed tests: GammaDistributionTest.testMath753Shape142:350->doTestMath753:297 shape = 142.0 , scale = 1.0 Old implementation ------------------ SummaryStatistics: n: 463 min: 0.0 max: 553.0 sum: 177972.0 mean: 384.3887688984882 geometric mean: 0.0 variance: 9958.419960169416 population variance: 9936.911493732765 second moment: 4600790.02159827 sum of squares: 7.3011228E7 standard deviation: 99.79188323791378 sum of logs: -Infinity New implementation ------------------ SummaryStatistics: n: 463 min: 0.0 max: 26.0 sum: 1139.0 mean: 2.460043196544277 geometric mean: 0.0 variance: 2.2013220760520977 population variance: 2.196567600725851 second moment: 1017.0107991360692 sum of squares: 3819.0 standard deviation: 1.483685302229586 sum of logs: -Infinity FastMathStrictComparisonTest.test1:90->setupMethodCall:205->callMethods:180->r eportFailedResults:156 double toRadians(-4.9E-324) expected -0.0 actual 0.0 entries [12] Tests in error: PSquarePercentileTest.testDistribution:755->doDistributionTest:712 ╗ OutOfMemory Tests run: 6193, Failures: 2, Errors: 1, Skipped: 37 >> >> >>> Phil >>>> --------------------------------------------------------------------- >>>> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >>>> For additional commands, e-mail: dev-h...@commons.apache.org >>>> >>>> >>> >>> --------------------------------------------------------------------- >>> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >>> For additional commands, e-mail: dev-h...@commons.apache.org >>> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >> For additional commands, e-mail: dev-h...@commons.apache.org >> >> > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
