All processes from the archive should be added. That is the intended behaviour. Can be related subprocesses or multiple processes related to one application. What is going wrong in your situation?
Second one does not happen if you use versioning IN the processdefinition afaik. Timestamps are dangerous for these.
Well, yes, of course all should be deployed, what happens, however, is that they are added tp each deployment which makes each appear n times.
The JBoss main deployer picks up the .bar, finds n .jpdl.xml files in it, goes and calls the jBPM deployer n times, which for the reason detailed in the OP will deploy all n processes for every call. Hence you get n x n processes.
This was clearly not how it worked before the quoted patch and I am not sure it is supposed to do so now.
The Deployer is still unde development. There were some issues to discuss and therefor it is in some unfinished state. I will come back to it asap, but don't know exactly when this will be :-(
The problem of not deploying it multiple times is not completly easy to solve, but I have some possible solutions in my mind and on an oldschool piece of paper in my office ;-)
Heh, OK, I know it's still being worked on but wanted to bring this specific point up. I first thought about filing a jira issue, but then decided against it and posted here instead.
FWIW, I worked around it in my local version by adding a flag to the addResourcesFromZipInputStream() method to tell it to NOT add any .jpdl.xml files. Then I included both calls in the deploy() sequence so it now adds all resources except the process descriptors and then adds the one process descriptor it has been invoked for.
Explicit versioning of the jpdl.xml files should prevent multiple deployments as I mentioned in my initial reply. Could very well be that that is not in place yet (and that Bernd has some ideas in this direction)
btw, *anything* not in the user guid but in the dev guide is still under development. So you could run into other issues as well as you use these