I do have an input dataset file, connecting to a Lookup as input link. This lookup has one more reference input, another dataset.
I lookup by one field, and set lookup stage on 'Continue' mode for unmatched records. On the lookup output, I do have all the columns for the upstream, plus one of the columns of the reference input (that goes out with a different name), set to be nullable (although in the reference input was not).
After that, I entered a Transformer to detect records that matched, and those did not. The idea is to have two outputs in the transformer:one for rows that matched, the other one for unmatched.
Let me try to paint it for you
Dataset2
|
|
|
|
DS1-----> Lookup ------> Transformer
To me, was clear that constraints for both output links were:
Code: Select all
IsNull(look1.COD_GR_EMP_FIC)
IsNotNull(look1.COD_GR_EMP_FIC)
After that, I changed the constraint to these functions that always worked for me in Server jobs
Code: Select all
len(trim(look1.COD_GR_EMP_FIC))=0
len(trim(look1.COD_GR_EMP_FIC))>0
Do you have any idea what all values are not null, even when it is proof than with the second method, there are rows coming with empty field, instead of null (unmatched records).
This turns to be the same when the column to check if there was a match is numeric. In these cases, value coming in the field is zero instead of null.
Thanks very much for your help