hi, everybody~
One side effect of the changeactivitystatebuilder method is that the current executionid is modified every time it is called.
Is there any way to modify the activity while keeping the current executionid unchanged?
No, currently not. The id should be internal anyway, what’s your use case?
When the process manually returns to the previous node, it needs to track the tasks in the same execution flow. If the execution-id changes, it will be difficult to track the execution track of the subsequent tasks. At the same time, it is not convenient to draw the process execution sequence diagram
In my opinion, if there is no parallel branch, the nodes in the same execution flow need to keep the execution ID unchanged when they jump freely
Parallel branch is one example, but there are many others where a new execution is created: subprocess, multi-instance, etc. So relying on the execution id wouldn’t help to cover all use cases. The proper way to implement this, is to rertrieve the BpmnModel, find the current activity and trace back through the sequence flow to what the previous activity was.
Your suggestion is very good, I will now find the relevant API to trace back from the model.
But I still hope that the simplest linear execution flow’s ChangeState can keep the exe-id unchanged. For this I always feel that the current processing scheme is a bit illogical. Or there may be a new version of the API that supports jump in single execution flow and keep the exe-id nochange.
Thank you very much.