After joining two datasets on A_Field1 = B_Field1, I want to check that A_Field2 = B_Field2. I don't need to include Field2 in the join as I will be dropping B_Field2 before the Writer. It's just to catch mismatches that may be down to typos, for example.
I can do this in Tester (A_Field2 = B_Field2), which correctly finds 2 mismatches out of 68,090 features.
In AttributeValidator, I thought this might work:
Attributes to Validate = A_Field2
Validation Rule = In
Rule Configuration = B_Field2
This matches 134 features, but rejects the rest. On the Readers, one field type is Integer while the other is Text. Could this explain it? But I thought the field types are effectively discarded once the data is read in...? And in any case, Tester works.
Am I doing something wrong in AttributeValidator, or is its behaviour different from Tester?