Giacomo Pati pisze: > > > Giacomo Pati wrote: >> I'm going to throw away my .m2, recompile it, and check your sample again. I >> just want to be sure, >> but this is going a little while. > > Ok, I've done as written above but have the same results. No matter what > settings I have in > core/cocoon-pipeline/cocoon-pipeline-impl/src/main/resources/META-INF/cocoon/spring/ObjectModel.xml > I always get the same stack traces: > > http://people.apache.org/~giacomo/GrekSample-pipelineComponent.log > http://people.apache.org/~giacomo/GrekSample-request.log > > Sorry, I get something different that you, so I'm stuck and cannot upgrade to > it with my projects!
Giacomo, this issue caused frustration that affected all of us but I would like to ask you about a little more. Could you start my sample in a debugger and set breakpoint on put() method of ObjectModelImpl? Then could you analyse all calls to that method and figure out if they are operating on the same ObjectModelImpl class instance? You should debug with OM's scope set to request. The expected result is that during whole request processing (including making a servlet call) every method call should be on the same instance of OM. If it's not the case just try to find out what and when creates new instances of OM. Since my sample works fine for Reinhard it must be something really unusual with your env. I'm going to describe in detail pipelineComponent scope as Daniel asked me for so it may be easier for you to track the code you are (hopefully) going to debug. -- Grzegorz Kossakowski Committer and PMC Member of Apache Cocoon http://reflectingonthevicissitudes.wordpress.com/