On 06/11/2012 01:49 PM, Bruce Dubbs wrote: > Fernando de Oliveira wrote: >> On 08-06-2012 03:21, Bruce Dubbs wrote:> I was able to build and test >> openjdk. The results are OK, but I'm >>> disappointed in the number of issues that are not addressed upstream. >>> >>> I used: >>> >>> #!/bin/bash >>> >>> export DISPLAY=:20 >>> Xvfb :20 -screen 0 1x1x24 -ac& >>> echo $!> Xvfb.pid >>> make jtregcheck -k 2>jdktest-errs |tee jdktest >>> kill -9 `cat Xvfb.pid` >>> unset DISPLAY >>> rm -f Xvfb.pid >> Don't you and/or DJ think these lines could be in the page? > Yes, I think they are useful, but I was waiting for DJ to have some time > to discuss. Yeah, but 1x1 session is not adequate for testing things like borders, window decorations, and transparent backgrounds, so I left it out for now. >>> There were several lines in jdktest-errs like >>> >>> WARNING: Path does not exist as file or directory: >>> Makefile:205: Extraneous text after `ifeq' directive >>> xxx uses or overrides a deprecated API >>> Note: Some input files use unchecked or unsafe operations. >>> >>> I just don't think a package as important as java should have those >>> types of errors/warnings in their test suite. >> Completely agree with this and the disappointment you mentioned in the >> first line above. > Since this was orifinally done by Sun, the tests probably assume a shell > program different from bash. That may explain some of the problems. > > Personally, I wouldn't release anything that causes any warnings. If > other environments cause warnings, then that is worth checking out and > either fixing the code or fixing the test.
With the patch I've just added, it stabilizes it a bit more, but there are still unexplained test failures if not in an absolutely pristine environment. I unfortunately don't have a better way to explain it in the book. All of the crypto and SSL failures are due to tests that have not been updated for newer NSS (padding failures). The JDK code has long ago been updated, but these tests are from a very old version of the full JTReg (as explained in the book). A 1x1 Xvfb session doesn't cut it either (see above), hence why I only added a mention of it instead of the actual instructions. As it is now, I am passing all remaining tests on i686 in TWM with the xset commands Bruce posted earlier. As I understand it, hacking up JTReg to work in the limited environment expected (no junit - and alot in there to assist) was a chore to begin with, and even more so with a newer version, so we simply have to wait or fix them ourselves and send upstream. IcedTea bugs #913 and #914 are to fix the NSS related failures, which will eliminate ~40 of the 80 (currently expected) failures, and are still slotted for the 2.2 release cycle (presumably icedtea-2.2.1). -- 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
