On Thu, Oct 16, 2008 at 9:38 AM, Dag H. Wanvik <[EMAIL PROTECTED]> wrote: > > I saw this interesting variant which indicates MessageBundleTest, but > the build succeeded. Maybe it is better to move this test to the > regression tests suite? > > : > junit-single: > [junit] Running > org.apache.derbyTesting.functionTests.tests.i18n.MessageBundleTest > [junit] Tests run: 1, Failures: 0, Errors: 1, Time elapsed: 0 sec > [junit] Test > org.apache.derbyTesting.functionTests.tests.i18n.MessageBundleTest FAILED > ***********************************************************************************^^^^^^^ > cleanjunit: > [delete] Deleting directory > /export/home/tmp/derby/sb/sb1/junit_20081016_1750 > > nojunittask: > > runmessagecheck: > > all: > > BUILD SUCCESSFUL > Total time: 1 minute 33 seconds > > khepri23:sb1$ echo $? > 0 >
It's interesting that this failed, I wonder why? This test was originally - a long time ago - part of derbyall, but was removed, and DERBY-1458 was logged to make it - or at least a check for the consistency of the messages - part of the build. But there were problems (see DERBY-1419). In a subsequent discussion, http://www.mail-archive.com/[email protected]/msg22107.html, it was decided it was better to have a build check... I guess we could add it into suites.All to be only run when running with classes, but I suspect that's not going to catch a lot of situations... Maybe I should modify the class to strip out the junit parts and just have a main and execute as a java program during the build? Myrna
