Detect deadlocking parallel gateway

[barmac]

The rule detects the following modeling patterns

• Detect when the number of incoming flows of a parallel gateway does not match the number of outgoing flows of the closest parallel gateway.

How does the rule improve the BPMN diagram?

The rule allows to prevent a deadlock.

Rule Details

• Name: no-parallel-gateway-deadlock
• Default notification level: error

What alternatives did you consider?

We could work on the name, but it's not a matter of style but rather the correctness of the diagram.

---

Context: https://camunda.slack.com/archives/C05FVD0AZ2R/p1731508704469399

[korthout]

My feeling is that this lint would have many false positives caused by cases with:

• forking behavior by having multiple outgoing sequence flows
• dynamic forking behavior by inclusive gateway
• dynamic forking behavior due to non-interrupting boundary events

It might be better to build this in a way starting with many false negatives and slowly build up to support more cases.

[nikku]

There is entire masterthesis about "livelyness detection", and it is not decisable in reasonable O(n) time. Let's be careful how/if we role this out. Also, let's carefully consider prior art.