For example, following issues seem to be persistently existing:
- Bug ID: 6566201 JNLP ClassLoader performance degraded in 1.6 significantly for the signed(?) jar
- Bug ID: 6532373 xcb_xlib.c:50: xcb_xlib_unlock: Assertion ’c->xlib.lock’ failed.
First issue means huge slow down in webstart application startup. When doing some profiling in webstart started application, it’s consuming huge amount of memory for storing CachedJarFile entries, where appearently major of the bulk is hanging in Manifest -entries, which are reduntantly kept in the memory. Great! This leakage is naturally not visible when starting same application normally, it’s visible only when starting application via webstart.
Second issue, is not actually specific to JDK6 (exists also in JDK5). But once again, look how nicely problem is bounced between java and open source geek developers. Both sides decline responsibility for the problem, and eventually nothing is fixed, except some dirty workarounds suggested. And appearently since some non-properly working workarounds exists…, whole issue is closed as ”fixed”. That ain’t proper.
Seemingly, this bug 6566201 is claimed to be fixed in JDK 1.6.0_10 release, which will be available by the summer 2008! Wow, I’m amazed due to so incredible fast response time for show stopper bug. Completely, utterly, ridiculous.
Currently this JDK 1.6.0_10 release is available via Early Access 6u10).