Great question. Cross-joins have the option to apply a filter to the results. The error indicates that the logical operation you specified for the filter function is failing with an error.
You should be able to get Xcalar to tell you the proper syntax and/or replicate the error by applying the filter operation directly on one of the constituent tables of the join.
The eval string for this filter would be:
If your filter example doesn't shed any light on the error, then the most common cause would be using a prefixed column in your operation (columns named, tablename::fieldname). You can make prefixed fields into derived fields by applying just about any modeling operation to the column.
For example, if you were trying to use the following field:
As the field is a string, but clearly contains integer content, you could even just select Change data type --> Integer from the column menu:
This operation will make the field into a derived field:
With the derived field, the syntax for the filter eval string becomes:
I would welcome feedback on how frequently members of the community run into this. Data prep inside Xcalar could make this an uncommon experience, but better to know for sure than to speculate.
Thanks for bringing this up, @sbrown!