Hi, this is a strange build bug. The only thing I see that could cause this is that with JDK 1.6 the classloader somehow clears the InlineEnclosureHandler class out of memory whereas in 1.7 it does not (tested against 1.7.0_21 too).
A possible fix could be using the test initialization to reset the counter to its default value (thus providing reproducible tests). I don't know how to do it properly without messing with visibility using Mockito (currently used in the project) but with PowerMock it would be something like WhiteBox.setInternalState(InlineEnclosureHandler.class, "counter", 0); I hope it helped, anyway I'm curious about the fix. Regards, __ Cedric Gatay (@Cedric_Gatay <http://twitter.com/Cedric_Gatay>) http://code-troopers.com | http://www.bloggure.info | http://cedric.gatay.fr On Sun, Apr 28, 2013 at 5:37 PM, Martin Grigorov <mgrigo...@apache.org>wrote: > Hi, > > Currently the build for master branch fails at BuildBot because of the > changes for https://issues.apache.org/jira/browse/WICKET-5085. > > It passes locally when using JDK 1.6.0_37 and fails when building with > 1.7.0_17. > If I run with -Dwicket.replace.expected.results=true then it fixes for 1.7 > and starts breaking for 1.6. > I have no idea why exactly this happens ... The InlineEnclosure id > generation now uses a _static_ counter. I see no reason why this static > variable has different values depending on the used JDK. > I'll continue debug it later today or tomorrow. > Any help is welcome! > > > -- > Martin Grigorov > Wicket Training & Consulting > http://jWeekend.com <http://jweekend.com/> >
