Olivier, You are right that variables are copied from the processinstance. I would have sworn that was only the case if it existed and if not, none was created. I'm almost positive of this. I'll look in the code but if the current behaviour is as you describe, boolean-expression: ((the docs have to be updated OR at least have to be more specific OR I have to read them again) AND there should be testcases for this) .... (I'm sure there were, but that could have been for 3.1.x, I'll check) but what about this scenario: I want to have the user enter a value for a variable, but if through any other means this variable was already set (e.g. the esb) I want to display that and the user should be able to change it. With the current behaviour and the way you have to configure that, there always is a variable either with the value null or the value that is already there. Then only putting a required=true on the input element in the webinterface would enforce this. In that case, I DO think there is a bug in the formgenerator since this is so unintuitive... I'm in favour of adding an additional check in the source then to see that if the current value is null and there is no value posted, an exception is thrown. Sigh...... View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4084098#... Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...