Niall Pemberton wrote at Montag, 12. Oktober 2009 23:33: > On Sat, Oct 10, 2009 at 4:10 PM, Oliver Heger > <oliver.he...@oliver-heger.de> wrote: >> Are these issues release blockers? >> >> I'd say if the 1.8.0 release had the same issues, they are not. Otherwise >> they should probably be investigated. > > I just tested with IBM JDK 1.6 using ant (m2 doesn't work for me with > that JDK) and the same test (MemoryLeakTestCase) fails in 1.8.0 as it > does in 1.8.1. > > As a quick recap - we changed the Map implementations in BeanUtils > 1.8.0 used in the static caches to be based on WeakHashMap because > using "class" as the key was preventing garbage collection of class > loaders in a web app environment resulting in a memory leak. This > resolves the memory leak issue in Sun's JDK implementations for JDK > 1.5+ (but not for JDK 1.3 or 1.4). Unfortunately it appears not to > work for IBM JDK 1.6. Why this is I don't have any idea at the moment. > > Since this isn't a regression in BeanUtils 1.8.1 (btw the memory leak > existed for all JDK implementations in BeanUtils 1.7.0) and I don't > have a fix for the IBM JDK I'm hoping that we can continue with this > release. If we find a solution that fixes it in the future, then I'd > be happy to do another release.
At least we should add it as known issue to the README. -Jörg --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org