Re: Testsuite changes this week
"Bryce W. Harrington" <[email protected]>
| Newsgroups | gmane.comp.lib.cairo |
|---|---|
| Message-ID | <[email protected]> |
git git git git 05/03 05/12 06/09 07/08 ----------------------------------------------- Tests run: 24838 24838 24813 24812 Passed: 15774 15743 16015 16001 Failed: 6476 6505 6226 6224 Expected Failed: 332 332 315 332 Error: 1 0 1 1 Crashed: 32 35 33 31 Untested: 2223 2223 2223 2223 Total: 24838 24838 24813 24812 The test suite triggered OOM killer on the system. I re-ran it on two other systems and it triggered OOM killer on those as well (one of the systems hung completely.) I'm not sure which test(s) did this though. Here's the tests that changed results between last month and now: FAIL -> PASS # degenerate-path.gl-window.argb32 FAIL -> CRASHED # pthread-same-source.svg11.argb32 CRASHED -> FAIL # pthread-same-source.svg11.rgb24 FAIL -> PASS # rel-path.gl-window.argb32 CRASHED -> FAIL # subsurface-modify-child.gl-window.argb32 CRASHED -> FAIL # subsurface-modify-child.gl-window-msaa.argb32 pthread always randomly crashes/fails, and looking at last month, the gl-window tests seem to do similarly. Perhaps the gl-window tests are the ones triggering OOM conditions? Memory leak somewhere? Anyway, the good news is that the past month's changes don't appear to have broken any tests. Bryce > Aside from downscale and pthread tests, these three tests also switched > from pass to fail on these two test runs: > > PASS -> FAIL # bitmap-font.gl-window.argb32 > PASS -> FAIL # degenerate-path.gl-window.argb32 > PASS -> FAIL # pass-through.gl-window.argb32 > > It's interesting all three are with the gl-window backend, but otherwise > the three tests have little commonality. It's not evident looking at > the test code why they'd be failing, and nothing in the git logs pops > out as directly relating to gl-window. Anyone else seeing these three > tests failing now, but passing as of a month ago? > > Bryce > -- > cairo mailing list > [email protected] > http://lists.cairographics.org/mailman/listinfo/cairo -- cairo mailing list [email protected] http://lists.cairographics.org/mailman/listinfo/cairo