yup, after reading your reply a couple of times i was able to decipher it :-)
A bit unfair when "random order" should suddenly be taken literally :-) I vaguely remember running into sometning similarly in the collections implementations on a project switching from JDK 5 to 6. The 2 tests should work now (I removed the uow.complete()). But probably the other tests in sample/dci should be scanned for similar potential issues too. /Kent On Tue, Sep 4, 2012 at 2:03 PM, Paul Merlin <[email protected]> wrote: > Kent, > > My last reply mentionned tests in sample/dci-cargo, yours mentionned tests > in sample/dci. > Sorry for my mistake. > > This change in JDK behavior is a mess. > In fact it's the previous behavior where getDeclaredMethods was documented > to return methods in random order but implementations were predictibles > that made developers used to the fact that tests are run in source order. > > We should welcome this change. > This means that non-independents tests need to be spotted and refactored. > > I ran the unit tests a lot of times while working on JDK7 support and > found bad tests in extensions/reindexer (my bad) and sample/dci-cargo but > the ones in samples/dci never failed here ... happy randomness ... > > There should not be a lot of non-independent tests anymore but we must be > vigilant. > > To be continued. > > /Paul > > > Kent Sølvsten a écrit : > >> sorry 'bout the broken commit. In my haste to flush everything from old >> laptop before switching day job, I must have made a partial commit >> somehow. >> >> I have problems with running the tests inside samples/dci when building on >> JDK7 on windows. The same tests are running fine on JDK6. >> >> It seems that for some obscure reason the test data (initialized inside >> @beforeClass method) are reinitialized in every test when using JDK7, >> which causes problems, since one of the tests removes money from a >> checking >> account, assuming that a previous test have moved money into that account. >> >> Any objections against me rewriting these tests to be completely isolated >> unittests, instead of depending on reusing data? >> >> /Kent >> >> >> >> >> On Tue, Aug 28, 2012 at 2:45 PM, Paul Merlin<[email protected]> wrote: >> >> Gang, >>> >>> Commit 55235f3c5194bf65a53b8f1abe2369****962019b9fa removed used files >>> and >>> so broken the build. >>> >>> See:https://github.com/Qi4j/**qi4j-**sdk/commit/**<https://github.com/Qi4j/qi4j-**sdk/commit/**> >>> 55235f3c5194bf65a53b8f1abe2369****962019b9fa<https://github.** >>> com/Qi4j/qi4j-sdk/commit/**55235f3c5194bf65a53b8f1abe2369**962019b9fa<https://github.com/Qi4j/qi4j-sdk/commit/55235f3c5194bf65a53b8f1abe2369962019b9fa> >>> > >>> >>> >>> I reverted this commit. >>> >>> >>> By the way, Qi4j now works with Java 7, the feature branch has been >>> merged. >>> >>> /Paul >>> >>> ______________________________****_________________ >>> qi4j-dev mailing list >>> [email protected] >>> http://lists.ops4j.org/****mailman/listinfo/qi4j-dev<http://lists.ops4j.org/**mailman/listinfo/qi4j-dev> >>> <http**://lists.ops4j.org/mailman/**listinfo/qi4j-dev<http://lists.ops4j.org/mailman/listinfo/qi4j-dev> >>> > >>> >>> ______________________________**_________________ >> qi4j-dev mailing list >> [email protected] >> http://lists.ops4j.org/**mailman/listinfo/qi4j-dev<http://lists.ops4j.org/mailman/listinfo/qi4j-dev> >> > ______________________________**_________________ > qi4j-dev mailing list > [email protected] > http://lists.ops4j.org/**mailman/listinfo/qi4j-dev<http://lists.ops4j.org/mailman/listinfo/qi4j-dev> > _______________________________________________ qi4j-dev mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/qi4j-dev
