Using CROSSFILTER to work around Single Direction Relationship

Anonymous 

I don't have time to optimize this measure... but you might not even need it. To use the measure correctly (and understand its behaviour) you have to have a correct dimensional model and follow Best Practices of dimensional modeling in PBI (for instance, you should never slice by attributes in the fact table apart from a degenerate dimension). It's up to you to create correct models and take precautions to guide users to use them correctly.

[First Isle Count] = 
var vAisleIsInScope = ISINSCOPE( WarehouseLocation[Aisle] )
var vCurrentAisle = SELECTEDVALUE( WarehouseLocation[Aisle] )
var vMissingAisleNotInScope = vCurrentAisle <> "missing"
RETURN
    IF( vAisleIsInScope && vMissingAisleNotInScope,
        
        // Need to find all the packages
        // in the current context that have
        // the selected isle as the first one.
        // This must happen after the removal
        // of all filters from WarehouseLocation
        // and exclusion of the missing isle. All
        // other filters ARE HONORED.
        
        CALCULATE(
            SUMX(
                DISTINCT( WarehousePackage[Package ID] ),
                1 * CALCULATE(
                    // Change MINX to MAXX to have the
                    // other measure, Last Isle Count.
                    vCurrentAisle = MINX(
                        SUMMARIZE(
                            WarehousePackage,
                            WarehouseLocation[Aisle]
                        ),
                        WarehouseLocation[Aisle]
                    )
                )
            ),
            REMOVEFILTERS( WarehouseLocation ),
            WarehouseLocation[Aisle] <> "missing"
        )
        
    )