Bugzilla – Full Text Bug Listing
| Summary: | test framework doesn't account for Java 8 VMs | ||
|---|---|---|---|
| Product: | [Eclipse Project] Platform | Reporter: | David Williams <david_williams> |
| Component: | Releng | Assignee: | David Williams <david_williams> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | CC: | jarthana, markus.kell.r, shankhba |
| Version: | 4.4 | ||
| Target Milestone: | 4.4 M7 | ||
| Hardware: | PC | ||
| OS: | Linux | ||
| Whiteboard: | |||
|
Description
David Williams
I'll take a crack at fixing logic to "work like it is supposed to", however, it has been "hacked" a few times in the past, and so I've no idea if it will really work any longer (without, major changes, in several areas). So, Plan B will be just to remove all the "restrictive" logic, and just run everything all the time. The logic was first added, I suspect, when we were sometimes using 4, sometimes 5, and then eventually sometimes VMs 1.6. These days, I can not imagine we'd ever run tests with anything lower than VM 1.6, so not sure all that "restrictive logic" is needed any longer? But, there might be special cases/platforms for which I'm unaware, plus, its hard to know how others might be making use of the test framework, so I hate to risk "breaking others" this late in the cycle, so Plan A will be to try and fix the logic to work as intended. This problem currently only shows up in our "Y-builds" ... such as see http://build.eclipse.org/eclipse/builds/4Y/siteDir/eclipse/downloads/drops4/Y20140316-0700/testResults.php The two "hyphens" in Linux column (currently using Java 8 on Linux only) mean the tests were not ran. (In general, the hyphen means "missing results" which could be caused by several issues). Since the "fix" for test framework is simply in a test.xml file, I can try to re-run that specific build's tests to see if my solution fixes it or causes other problems ... which means at some point the current "test results" will be replaces by new results (either better or worse, depending on if my fix works, or makes things worse :) Proposed fix: http://git.eclipse.org/c/platform/eclipse.platform.releng.aggregator.git/commit/?id=cd9e7ae3ce884aacd81abf50b472eac93e52279f apologies for all the formatting changes .... but, had to do most of that just to make sense of it all. See also bug 429571 comment 27. Could just be a matter of setting property "J2SE-5.0" (and J2SE-7.0?) again. Can just point to the Java 8 VM. E.g. in I20140311-1200, linux.gtk.x86_64_6.0_consolelog.txt contained: [echoproperties] J2SE-5.0=/opt/public/common/jdk1.7.0_51/jre/bin/java [echoproperties] J2SE-6.0=/shared/common/jdk1.7.0-latest/jre/bin/java [echoproperties] J2SE-7.0=/opt/public/common/jdk1.7.0_51/jre/bin/java (In reply to Markus Keller from comment #4) > See also bug 429571 comment 27. Could just be a matter of setting property > "J2SE-5.0" (and J2SE-7.0?) again. Can just point to the Java 8 VM. OK, comment 3 looks like it does exactly that. I missed that the properties are actually set in the same script. Re-test looks good for Linux. While still running, I checked Hudson and this (short lived) URL shows one of the missing tests results: https://hudson.eclipse.org/hudson/job/ep44Y-unit-lin64/ws/workarea/Y20140316-0700/eclipse-testing/results/html/org.eclipse.pde.api.tools.tests_linux.gtk.x86_6.0.html So, hopefully I didn't break "other" platforms when Java 8 not present ... but ... given the time ... will find out at 7 PM Y build. *** Bug 430869 has been marked as a duplicate of this bug. *** |