We want to create condition expressions on data that is not stored in task variables but in our own datastore. So we need to influence the way variables are resolved (this was perfectly ok how it worked in jbpm3: extending JbpmVariableResolver and retrieving the variables from our datarepository) Another option would be to create a java class which can handle some kind of expression language to generate a transition name, but I have a feeling this is duplicating code (juel is fine for us) and this also eliminates the possibility to use decision nodes (we have to change them to java nodes), which influences the process drawing in a way that it is not really correct. Or is there some other solution you can think of? I liked the solution we did for jbpm3, that is way I was wondering if something similar will be created in jbpm4, but maybe there is another solution in jbpm4? View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4250535#... Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&a...