Support for multi instance tasks
g-zimmer opened this issue · comments
The new data flow analysis feature does not consider multi instance tasks/subprocesses, which loop over collections. See https://docs.camunda.org/manual/7.11/reference/bpmn20/tasks/task-markers/
... and you are using those in your project, @g-zimmer?
Yes, actually quite frequently. In my current customer project, every single process contains multiple multi instance tasks.
If we realize this feature, it should provide the following features:
- Analyze expressions for read operations in variable ‘loopCardinality’, if attribute is specified (e.g.
<loopCardinality>${nrOfOrders-nrOfCancellations}</loopCardinality>
references undefined variables). - Read-Operations for the local variables
nrOfInstances
,nrOfActiveInstances
,nrOfCompletedInstances
andloopCounter
do not lead to a UR anomalies. - If a collection (attribute ´loopDataInputRef´) is specified, the referenced variable must exist (i.e. prevent UR anomalies, if referenced variable does not exist). Read operations for specified
element variable
(attributeinputDataItem
) do not lead to an UR anomaly.
@g-zimmer : would this fit your needs?
@TamaraGunkel : could you analyze, if these requirements could be realized with relatively low effort (e.g. by adapting input/output mappings)?
@cl-a-us : yes, that would fit our needs (particularly the collection variable check, as we usually iterate over collections).
Yes, I think if we're using a similiar construct like i/o mapping, it should be relatively easy to implement. I'll look at this now
The feature is now included in the development branch.
@g-zimmer Can you check if it satisfy your needs?