Hi, Those classes are in wicket jars but if they were not loaded by the WebappClassLoader before the call to javax.servlet.Filter#destroy() then the web container won't be able to load them inside #destroy(). It is some kind of security protection done by the web container to not load any new classes once destroy is initiated.
The usual solution is to pre-load (at Application#init()) all classes which might be needed in Application#destroy() but this might break at any time when some new code is added to #destroy(), even in the application code - Application#onDestroy(). On Thu, Nov 29, 2018 at 10:15 AM Rob Audenaerde <rob.audenae...@gmail.com> wrote: > Hi all, > > The problem where WicketApplication.destroy() hangs in 8.1.0 seems to be > fixed with WICKET-6603, however, I noted a new problem (less impacting for > me at least). > > When quickly firing up embedded jetty, then opening a page and shutting > jetty down, I get exceptions in the page serializer. Note that this has > nothing to do with the WicketTester, as I use a pre-compiled > application.war in a standalone embedded Jetty. > > (Jetty version: 9.4.11.v20180605, Java version: build > 10.0.2+13-Ubuntu-1ubuntu0.18.04.4, wicket version 8.2.0) > > I got at least these two on different occasions: > > Exception in thread "Wicket-AsyncPageStore-PageSavingThread" > java.lang.NoClassDefFoundError: > org/danekja/java/util/function/serializable/SerializableBiConsumer > at java.base/java.lang.Class.getDeclaredMethods0(Native Method) > at java.base/java.lang.Class.privateGetDeclaredMethods(Class.java:3119) > ... > > and > > Exception in thread "Wicket-AsyncPageStore-PageSavingThread" > java.lang.NoClassDefFoundError: > org/apache/wicket/request/mapper/parameter/INamedParameters$Type > at java.base/java.lang.Class.getDeclaredMethods0(Native Method > ... > > There classes are present in the wicket jars, so I'm a bit at loss as to > what causes this. I can simple wrap the `destoy()` of my application in a > try-catch and ignore the exceptions; however; it seems something is still a > bit wrong. > > It seems these classes are unloaded somehow? Maybe any of you can shed some > light on this? > > -Rob >
