Error "rules is cyclic or exceeds maximum cascading level"

Zoggo

I'm using a Attribute Calculation  Rule in ArcGIS Pro 3.0.2. It's a line featureclass in a fgdb. I checked "Exclude from application evaluation and triggered Insert und Update. The following very simple code is working when I alter the geometry:

var feature_ = $feature; 
return {
    "edit": [{
        "className": "HydroEdgePro",
        "updates": [{
            "objectID": feature_["OBJECTID"],
            "attributes": {
                "BEMERKUNG": "test 2"
            }
        }]
    }, {
        "className": "arcade_log",
        "adds": [{
            "attributes": {
                "Value": "test"
            }
        }]
    }]
}

  

If I try to do something with the $FeatureSet (see example below) then I always get the Error "The evaluation of attribute rules is cyclic or exceeds maximum cascading level.":

var feature_ = $feature;

var features = $FeatureSet;
var test = Count(features); 

return {
    "edit": [{
        "className": "HydroEdgePro",
        "updates": [{
            "objectID": feature_["OBJECTID"],
            "attributes": {
                "BEMERKUNG": "test 2"
            }
        }]
    }, {
        "className": "arcade_log",
        "adds": [{
            "attributes": {
                "Value": "test"
            }
        }]
    }]
}

I don't understand why this happens. During debug mode the code is working.

 

HusseinNasser2

When you create a feature in the geodatabase it goes into a memory buffer for performance reasons. Additional updates to the feature by its object id will go to the buffer and then finally flushed at one swoop to the database. 

This is what happened in your first rule, you created the feature, then asked AR to update the same feature field BEMERKUNG to "Test 2". The AR found that the feature is still in memory and updated the field in memory then flushed it to the database. So the database only saw one insert. 

 

However we can't always do this optimization. This optimization is not performed when a query against the class happens, because the query must see the feature being created (basic database semantics), so when a query happens to the class (that is your second rule where you did a count of the class), we had to flush the inserted feature to the database. Then the AR says wait a minute, we need to update the same feature (that is your return dictionary with objectId), that update actually becomes a physical second update which triggers the attribute rule again, and we go into the infinite cycle. 

 

So as a better practice, the attribute rule need to be written in a predictable way, if you want to updates fields in the same feature use the return "result" dictionary, with this we have knowledge in the AR to update the same row.  

Hope that clarifies it.