Hi Jörg, Am 10.01.2014 18:04, schrieb Jörg Schaible: > Hi Oliver, > > Oliver Heger wrote: > >> Hi Jörg, >> >> Am 10.01.2014 01:01, schrieb Jörg Schaible: >>> Hi Oliver, >>> >>> Oliver Heger wrote: >>> >>>> Hi Jörg, >>>> >>>> thanks for your review. >>>> >>>> Am 08.01.2014 10:55, schrieb Jörg Schaible: >>>>> Hi Oliver, >>>>> >>>>> README.txt is badly out of date, since it claims we ship >>>>> commons-beanutils, commons-beanutils-core and >>>>> commons-beanutils-bean-collection. >>>> Good spot. I think the file can be removed completely. I will do this in >>>> trunk, but I hope there is no need to create another RC. The situation >>>> with the collections jars is also explained on the main site. >>>> >>>> Regarding the failing test cases: Most of them seem to be related to >>>> memory leak tests. I guess, here a special strategy is used which may >>>> well depend on the JDK. However, the tests affected have not been >>>> touched recently - certainly not for the 1.9.1 release and probably not >>>> for 1.9.0. So I assume the problems already existed in earlier releases. >>> >>> I had no time to run my tests for 1.9.0, but there were no errors for >>> 1.8.3. Were those tests new? >> >> No, these tests are not new. And there were no functional changes on >> them (they may have been touched for the 1.9.0 release to fix warnings >> related to generics). >> >> Are you sure that they were successful for 1.8.3? > > I found my VOTE for 1.8.3 in the archives and I am quite sure I ran my > standard procedure ;-) > > The IBM JDKs might have had a lower minor version but existed. Strange, > indeed.
Yes. I had another look at the affected tests and their history and did not find any significant changes in the way they work. Also, according to the release notes, there are no changes in the caching behavior in [beanutils] since the 1.8.x releases. During the vote for this release, some reviewers also reported test failures with the MemoryLeakTestCase class in some environments (which I could not reproduce). Sebb was able to fix the continuum build by increasing the maximum memory size in the surefire configuration. My guess is that these memory leak tests have been fragile from the very beginning. Maybe a change in an internal data structure is sufficient to let them fail in certain environments, because now more or less memory is used. So IMHO the problem is on the side of the tests and not in the implementation. Therefore, I am going to continue with the release; I am pretty sure that 1.9.1 does not make the situation worse compared to 1.9.0. Another argument is that the tests are successful in all newer JDK implementations. Thanks again for running all these tests! Oliver > > Cheers, > Jörg > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
