It must be the 'jvm' property in that file. When I add that, it uses the Harmony JRE to run the tests. However, the test harness is still run with the public JRE (in my case, Sun's JRE).
Would you mind trying to add "-vm path/to/harmony" to runtests and seeing what results to get from a clean start? I'd like to know if you're getting the same failures. The console output doesn't show much, but when you look at the results, no tests were run and the log shows some weird errors in equinox. -Nathan On Mon, Jan 12, 2009 at 12:47 AM, chunrong lai <[email protected]> wrote: > I run it with > runtests -properties eut.run.properties jdtcorebuilder > > or > D:\eclipse-testing>java -jar > eclipse\plugins\org.eclipse.equinox.launcher_1.0.200.v20080825-1800.jar > -application org.eclipse.test.coretestapplication -data workspace -dev bin > formatter=org.apache.tools.ant.taskdefs.optional.junit.XMLJUnitResultFormatter,d:\eclipse-testing\eclipse\org.eclipse.jdt.core.tests.builder.BuilderTests.xml > -testPluginName org.eclipse.jdt.core.tests.builder -className > org.eclipse.jdt.core.tests.builder.BuilderTests -os win32 -ws win32 -arch > x86 > after manual clean/extraction. > > I also tried "runtests -properties eut.run.properties" and saw those > jdtcorebuilder errors are reproducible. > On Mon, Jan 12, 2009 at 12:42 PM, Nathan Beyer <[email protected]> wrote: > >> When I don't specificy "-vm", then it uses the 'java' executable in my >> path that Sun's public JRE installed, for all invocations. All I have >> set in my properties file is the "J2SE-1.5" property. What does the >> "jvm" property do? >> >> I presume you're running "runtests -properties eut.run.properties all". >> >> -Nathan >> >> On Sun, Jan 11, 2009 at 10:19 PM, chunrong lai <[email protected]> >> wrote: >> > hi, Nathan: >> > I have not tried 3.5M4 but just tried 3.5M3. >> > According to the attched "Readme.htm", one only needs to specify the >> > (J2SE) item of eut.run.properties (I attached mine) and then run >> > >> > runtests -properties <path to properties file> [-os <operating system>] >> [-ws >> > <windowing system>] [-arch <architecture>] [-noclean] [-vm <path to java >> > executable>][-Dtest.target=performance][targetname1 targetname2 >> targetname3 >> > etc...] >> > for that. You can read the "Requirements" session in it. It seems >> that >> > the requirements (winzip, ant) are minimum. >> > Detailed steps of the testing can be found in >> > http://www.eclipse.org/articles/Article-PDE-Automation/automation.html(around >> > Fig 13). >> > On Mon, Jan 12, 2009 at 11:48 AM, Nathan Beyer <[email protected]> >> wrote: >> >> >> >> I've been trying to recreate these errors, can anyone share the >> >> specific commands, settings and tools being used? >> >> >> >> I'm trying this with the 3.5M4 builds and I can't even get the 'ant' >> >> test to run when Harmony is the bootstrapping JRE and the JRE running >> >> the actual tests. >> >> >> >> -Nathan >> >> >> >> On Sat, Jan 10, 2009 at 7:35 AM, Gregory Shimansky >> >> <[email protected]> wrote: >> >> > On 7 January 2009 Huang, Eric wrote: >> >> >> Did not cath this one earlier. What is the jrt.core.tests.builder >> >> >> about? It >> >> >> almost failed completely. >> >> > >> >> > It looks like almost all tests fail with the same exception message: >> >> > >> >> > Build path contains duplicate >> >> > entry: >> >> > >> 'D:harmony.cc/trunk/working_vm/build/windows_x86_msvc_release/deploy/jdk/jre/bin/default/gc_gen.jar' >> >> > for project 'Project' >> >> > >> >> > Interesting how there is no '/' or '\' after D: in the path name. Is >> it >> >> > a >> >> > valid path in win32? >> >> > >> >> >> Eric >> >> >> >> >> >> -----Original Message----- >> >> >> From: chunrong lai [mailto:[email protected]] >> >> >> Sent: Wednesday, January 07, 2009 12:32 PM >> >> >> To: [email protected] >> >> >> Subject: Re: testing for the newer EUT >> >> >> >> >> >> After works on HARMONY-6020 now we get 98.42% pass rate for 26416 >> test >> >> >> cases when running EUT35 in win32 platform as below. >> >> >> >> >> >> Tests >> >> >> Failures Errors Success rate >> >> >> org.eclipse.ui.tests 2143 >> >> >> 19 1 99.07% >> >> >> org.eclipse.jdt.core.tests.model 11138 >> >> >> 0 6 99.95% >> >> >> org.eclipse.jdt.core.tests.compiler 10216 >> >> >> 17 0 99.83% >> >> >> org.eclipse.jdt.core.tests.builder 195 >> >> >> 0 184 5.64% >> >> >> org.eclipse.jdt.core.tests.performance 1 >> >> >> 0 0 100.00% >> >> >> org.eclipse.team.tests.core 17 >> >> >> 0 0 100.00% >> >> >> org.eclipse.update.tests 131 >> >> >> 0 0 100.00% >> >> >> org.eclipse.core.expressions.tests 98 >> >> >> 0 0 100.00% >> >> >> org.eclipse.osgi.tests 504 >> >> >> 48 101 70.44% >> >> >> org.eclipse.core.tests.resources 914 >> >> >> 7 29 96.06% >> >> >> org.eclipse.core.tests.runtime 389 >> >> >> 1 0 99.74% >> >> >> org.eclipse.compare.tests 80 >> >> >> 0 0 100.00% >> >> >> org.eclipse.ant.tests.core 85 >> >> >> 0 0 100.00% >> >> >> org.eclipse.releng.tests 5 >> >> >> 0 0 100.00% >> >> >> org.eclipse.jdt.text.tests 500 >> >> >> 0 0 100.00% >> >> >> >> >> >> Totally: 26416 >> >> >> 92 321 98.42% >> >> >> >> >> >> More details are posted in HARMONY-6062. >> >> >> >> >> >> regards, >> >> >> chunrong >> >> >> Managed Runtime Technology Center, Intel >> >> >> >> >> >> On Sat, Dec 27, 2008 at 3:25 AM, Nathan Beyer <[email protected]> >> >> >> wrote: >> >> >> > I posted a comment about this on the JIRA issue - I think this is >> >> >> > related to a class loading problem I found a few weeks ago. The >> >> >> > ClassNotFoundError looked familiar - I reduced it to a complex >> >> >> > cyclical class dependency that DRLVM couldn't load. >> >> >> > >> >> >> > https://issues.apache.org/jira/browse/HARMONY-6020 >> >> >> > >> >> >> > On Thu, Dec 25, 2008 at 10:32 PM, chunrong lai >> >> >> > <[email protected]> >> >> >> > >> >> >> > wrote: >> >> >> > > hi, colleagues: >> >> >> > > I'd like to update my running status of EUT35 with you. As >> >> >> > > reported >> >> >> > > in HARMONY-6062, I have not been able to run EUT35 with Harmony >> >> >> > > (although it >> >> >> > >> >> >> > is >> >> >> > >> >> >> > > fine to run EUT33 with it). >> >> >> > > The failure can be reproduced with the long java command line >> >> >> > > option >> >> >> > > "java -jar >> >> >> > > >> >> >> > > >> eclipse\plugins\org.eclipse.equinox.launcher_1.0.200.v20080825-1800.jar >> >> >> > > -application org.eclipse.test.uitestapplication -data workspace >> >> >> > > -dev >> >> >> > > bin >> >> >> > >> >> >> > >> >> >> > >> formatter=org.apache.tools.ant.taskdefs.optional.junit.XMLJUnitResultForm >> >> >> >> >> >> > >> >atter,d:\eclipse-testing\eclipse\org.eclipse.jdt.text.tests.JdtTextTestSui >> >> >> >te.xml >> >> >> > >> >> >> > > -testPluginName org.eclipse.jdt.text.tests -className >> >> >> > > org.eclipse.jdt.text.tests.JdtTextTestSuite -os win32 -ws win32 >> >> >> > > -arch >> >> >> > >> >> >> > x86". >> >> >> > >> >> >> > > I also attached the output file in the JIRA, where I guess the >> root >> >> >> > > cause are from some Harmony jar file constrains, "!MESSAGE >> Missing >> >> >> > > Constraint: Bundle-RequiredExecutionEnvironment: JavaSE-1.6". >> >> >> > > Another interesting point is that after the test case had >> >> >> > > successfully runed (so the configuration phase has been finished) >> >> >> > > with >> >> >> > > RI (with above command), it would be no problem for Harmony to >> run >> >> >> > > it >> >> >> > > again. This >> >> >> > >> >> >> > somehow >> >> >> > >> >> >> > > supports our guess of some verification failures in the >> >> >> > > configuration >> >> >> > >> >> >> > phase. >> >> >> > >> >> >> > > On Thu, Dec 18, 2008 at 8:37 PM, chunrong lai >> >> >> > > <[email protected]> >> >> >> > >> >> >> > wrote: >> >> >> > >> I am running the tests, not finished yet. >> >> >> > >> >> >> >> > >> Thanks. >> >> >> > >> >> >> >> > >> >> >> >> > >> On Wed, Dec 17, 2008 at 9:28 PM, Tim Ellison >> >> >> > >> <[email protected] >> >> >> > > >> >> >> > >wrote: >> >> >> > >>> Are you looking at this Chunrong? >> >> >> > >>> >> >> >> > >>> Tim Ellison wrote: >> >> >> > >>> > chunrong lai wrote: >> >> >> > >>> >> So far Harmony has only EUT3.3 in testing. It looks like >> it >> >> >> > >>> >> is >> >> >> > >> >> >> > time >> >> >> > >> >> >> > >>> to >> >> >> > >>> >> >> >> > >>> >> include EUT3.4 now. Your suggestions/comments are welcome. >> >> >> > >>> > >> >> >> > >>> > Skip 3.4 and go straight to the 3.5 stream tests [1]. >> >> >> > >>> > >> >> >> > >>> > It would be prudent to test using the latest milestone build >> >> >> > >>> > first >> >> >> > >> >> >> > [2], >> >> >> > >> >> >> > >>> > but if we want to remain current then catching any issues >> >> >> > >>> > earlier >> >> >> > >>> > in >> >> >> > >> >> >> > the >> >> >> > >> >> >> > >>> > integration builds would be the thing to do. >> >> >> > >>> > >> >> >> > >>> > How about running then using the Apache infrastructure, which >> >> >> > >>> > would >> >> >> > >> >> >> > give >> >> >> > >> >> >> > >>> > us better control of the jobs and insight into the failures? >> >> >> > >>> > >> >> >> > >>> > [1] http://download.eclipse.org/eclipse/downloads/ >> >> >> > >>> > [2] >> >> >> > >> >> >> > >> >> >> > >> http://download.eclipse.org/eclipse/downloads/drops/S-3.5M3-200810301917/ >> >> >> >index.php >> >> >> > >> >> >> > >>> > Regards, >> >> >> > >>> > Tim >> >> > >> >> > >> >> > >> >> > -- >> >> > Gregory >> >> > >> > >> > >> >
