Calculating new column using row values from same table: circular dependency; avoiding EARLIER?

Hello,

I have a LogTable listing all mutation events on some objects.

The row format is:

Object_ID: the object on which mutations are carried on

Mutation_Date: Timestamp of the mutation

Mutation_Parameter: the object parameter on which the mutation is done, e.g. Status, Urgency, Other

Old_Value: the value of the object parameter prior to the mutation

New_Value: the value of the object parameter after the mutation.

I would like to create a new column "Status_before_Mutation" showing the value of the status parameter prior to each mutation. In case of a status mutation, this is logically equal to the existing value in column Old_Value, but in case of other mutations, I need to consider the latest change of status prior to the time of the mutation being considered.

For each mutation, the logic I am pursuing is to select all rows with the same Object_ID, then filter rows with MutationParameter = "Status", then filter rows with Mutation_Date prior to Mutation_Date of the mutation being considered, sort the remaining rows in descending order of Date, pick the first one and extract the column vale "Old_Value" from this row.

Unfortunately, I get a circular dependency error. I guess I need to exclude the new column "Status_before_Mutation" from the CALCULATETABLE row context, which I thought I did with the ALLEXCEPT... 

Here is the code leading to this circular dependency: 

Status_before_Mutation = 
    CALCULATE(
        VALUES(LogTable[Old_Value]);
        TOPN(
            1;
            CALCULATETABLE(
                LogTable;
                ALLEXCEPT(LogTable; LogTable[Object_ID]);
                LogTable[Mutation_Parameter] = "Status";
                LogTable[Mutation_Date] <= EARLIER(LogTable[Mutation_Date])
            );
            LogTable[Mutation_Date];
            DESC
        )
    )              

Another potential issue is maybe the use of EARLIER, which I have read is not recommended for large tables.

The LogTable is quite big, up to several hundred thousands rows. I have several thousands different objects each with a number of mutations in the range [1...50].

Any idea on solving this, possible avoiding the use of EARLIER?

Thanks for any suggestion.

Karim