Hi Ive,

With a custom operator it is currently not possible to access all column values at once. However, it should still be possible to apply any sort order.

The key is to find a mapping of for the cell values you want to sort to a related set a values that can then be sorted by default sort operator.

This mapping can be implemented using a custom operator, and can be linked to your original column for sorting by using the Sort Redirector operator from the catalog.

Example

Let's assume you have a column called "Numbers" with possible values "One", "Two", "Three", "Four" that when sorted should be in this numerical order.
By default, when sorting on this column the values would be sorted alphabetically: "Four", "One", "Three", "Two". We don't want this.

To change the sort order:

1.  Create a new custom operator "Calculate sort values" that does the following:
   1. Takes as input the column with the original values
   2. Adds a new integer column, let's call it "Sortable"
   3. For every row in the data set:
      1. Gets the original cell value from the row
      2. Maps the original value to its numeric value ("One" -> 1, "Two" -> 2 etc.)
      3. Stores the numerical value in the new "Sortable" column of this row
2. Create the following query:
   1. Your original query
   2. Custom operator: "Calculate sort values"
      1. Argument: "Numbers" column
   3. Custom operator: Sort Redirector
      1. Redirect from column: "Numbers" column
      2. Redirect to column: "Sortable" column
   4. Select: all but the "Sortable" column to remove it from the end result

Now, whenever a Sort operation is performed on the "Numbers" column of this query, the rows will be sorted using the corresponding numerical values from the "Sortable" column.

Note

As of DataMiner 10.5.9, Custom Operators do not support real-time updates yet. If this is a requirement, the only viable solution is consolidate all the custom operator logic together with the rest of your query into a single ad hoc data source.