On 06/02/2012 09:32 PM, DJ Lucas wrote: > Okay, so digging into the test results on the new OpenJDK/IcedTea, the > following jdk test failures can be ignored while testing on icedtea-2.2: > > Just append the following to icedtea-2.2/test/jtreg/excludelist.jdk.jtx > to actually skip the tests > <Snip>
In addition to the above, I've found that RedHat is using Xvfb to run the display tests... export DISPLAY=:20 <http://pkgs.fedoraproject.org/gitweb/?p=java-1.7.0-openjdk.git;a=blob;f=java-1.7.0-openjdk.spec;h=9bfd2949fb14ac75590148a3eb064b90166081a3;hb=f2f1a7cb522137f2078b9a75388efa7c53fce8df#l918>Xvfb :20 -screen 0 1x1x24 -ac & echo $! > Xvfb.pid make jtregcheck -k <http://pkgs.fedoraproject.org/gitweb/?p=java-1.7.0-openjdk.git;a=blob;f=java-1.7.0-openjdk.spec;h=9bfd2949fb14ac75590148a3eb064b90166081a3;hb=f2f1a7cb522137f2078b9a75388efa7c53fce8df#l921>kill -9 `cat Xvfb.pid` <http://pkgs.fedoraproject.org/gitweb/?p=java-1.7.0-openjdk.git;a=blob;f=java-1.7.0-openjdk.spec;h=9bfd2949fb14ac75590148a3eb064b90166081a3;hb=f2f1a7cb522137f2078b9a75388efa7c53fce8df#l922>unset DISPLAY <http://pkgs.fedoraproject.org/gitweb/?p=java-1.7.0-openjdk.git;a=blob;f=java-1.7.0-openjdk.spec;h=9bfd2949fb14ac75590148a3eb064b90166081a3;hb=f2f1a7cb522137f2078b9a75388efa7c53fce8df#l923>rm -f Xvfb.pid ...that handles the screensaver and screen blank settings, and gets us away from any WM discrepancies and user interaction. I'm not so sure about 1x1 but if it works, great. I'm going to give that a shot with the tests removed for WM compat put back and see where we are. Hopefully I'll have the book updated later this afternoon with whatever results I have. I have come to the conclusion that the test suite is just too fragile to be run on a system that is doing anything else, though it is much improved over the ~180 failures in the IT6 version. My plan for the book is for any test that I've restored that still results in failure to be removed again, as well as the NSS dependent tests that seem to still be broken, and leave the current note with an estimated 60 failures since my results are fairly consistent on both supported arches. With the other 17 removed, I'm down to zero failures on both if I do nothing, but I'm also in a very limited, and dedicated environment. Test time outs occur while copying the resultant j2sdk out of the source tree for instance. While Xvfb does give some of that back, I'm not convinced that the suite itself is ready for a perfect run by an end user, there are just too many environmental factors to consider (nor would I expect a user to adhere to such a limiting set of expectations). -- DJ Lucas -- This message has been scanned for viruses and dangerous content, and is believed to be clean. -- http://linuxfromscratch.org/mailman/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
