Hi Corticon community!

I’m Mrak Pousee, and I’m diving into a bit of a puzzling issue with the Corticon 7 rule enginespecifically around how Shared Data Objects (SDOs) are updated and reused across rule execution cycles.

Here’s what’s going on:

• I defined an SDO (say, CustomerOrder) with a nested collection of LineItems.
• In the first execution pass, the rules evaluate correctly, and the SDO gets updated as expected.
• However, in subsequent runs—even when invoking the same ruleset with new input—the SDO seems to retain stale data in LineItems, resulting in unexpected aggregations and logic misfires.
• I’ve tried isolating the issue by resetting the session context and forcing the SDO to reinitialize, but it seems like the engine isn't flushing the prior state as intended.

Here’s what I’ve already checked:

• Ensured I'm using a fresh Process Session for each execution.
• Used explicit Session.invalidate() to clear context—still no luck.
• Checked that the rule task definition doesn’t include any caching flags that could preserve state across runs.

So I’m wondering:

1. Is this a known quirk in Corticon’s handling of SDOs when data structure definitions remain constant across runs?
2. Does anyone use a custom pattern to truly clear or clone SDOs between executions?
3. Are there recommended practices—like forcing deep copies or leveraging specific session constructs—to avoid this kind of state carryover?

I’m hoping to make each rule run purely stateless, and any insight, workarounds, or best-practice examples would be immensely helpful.

Thanks so much for any tips or code snippets you can share!

Mark Pousee