Bugzilla – Full Text Bug Listing
| Summary: | Remove performance target from "osgi tests" | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Eclipse Project] Equinox | Reporter: | David Williams <david_williams> | ||||
| Component: | Framework | Assignee: | equinox.framework-inbox <equinox.framework-inbox> | ||||
| Status: | RESOLVED FIXED | QA Contact: | |||||
| Severity: | normal | ||||||
| Priority: | P3 | CC: | tjwatson | ||||
| Version: | unspecified | ||||||
| Target Milestone: | Neon M4 | ||||||
| Hardware: | PC | ||||||
| OS: | Linux | ||||||
| Whiteboard: | |||||||
| Bug Depends on: | |||||||
| Bug Blocks: | 454921 | ||||||
| Attachments: |
|
||||||
I think at status meeting you mentioned "no longer relevant" since they are for a "resolver" we no longer use ... and if I heard that right, thought I'm mention it'd still be valid (and even good) if you could come up with a "resolver performance test" that was independent of any specific implementation ... but not sure if that's possible, or perhaps that's already tested indirectly in other ways. (In reply to David Williams from comment #1) > I think at status meeting you mentioned "no longer relevant" since they are > for a "resolver" we no longer use ... and if I heard that right, thought I'm > mention it'd still be valid (and even good) if you could come up with a > "resolver performance test" that was independent of any specific > implementation ... but not sure if that's possible, or perhaps that's > already tested indirectly in other ways. You are correct, we should do this. Will take time to do. Basically would involve creating a Equinox instance and blasting it with a large number of bundles that represent the resolution scenarios that we want to do performance tests on. Were these never commented out? Of the test.xml file? I've not seen any "osgi tests" test results BUT, I still them "installed" in the log. I think we are just wasting time, installing a fresh "eclipse", installing the OSGi tests (for which there are none) and then moving on to the next suite, which requires deleting the previous installation, and then installing a fresh version again. If there are some that are supposed to be running, then they are broken. If there are none, then please remove the target from the test.xml file so we won't waste time on them. Let me know. FYI, this is the director log where I can tell the performance tests are still enabled in the test.xml file, but ... there are not results, no logs, etc. https://hudson.eclipse.org/perftests/view/Eclipse%20and%20Equinox/job/ep46I-perf-lin64/ws/workarea/I20151103-0800/eclipse-testing/results/linux.gtk.x86_64_8.0/directorLogs/director-org.eclipse.osgi.compatibility.plugins.feature.feature.group.log This is the only mention, in the general log: [echoproperties] org.eclipse.osgi.tests=org.eclipse.osgi.tests_3.11.0.v20151020-1445 [echoproperties] org.eclipse.osgi.tests.has.performance.target=true |
Created attachment 246681 [details] patch to comment out perf tests in osgi.tests As mentioned at status call, the current "osgi performance tests" are not relevant on current code stream, and could be removed. I can certainly remove them from my "test framework" list ... but, ideally, someday, I won't need a test framework list, and can just run anything that is marked with "performance" target. Therefore, I recommend these have their target removed. You can comment them out, as I've done in this patch, or just literally remove that particular code from test.xml files (and perhaps even the the code that does the tests! If you really wanted to stay "clean"). As mentioned, that's not to say there are not other "osgi performance tests" but not sure if they'd be ran directly here or via some other suite. Thanks,