Join Failure

[NBPSE]

I have two inputs to my join. Some months both inputs have data, some months only 1 has data. On the months only 1 has data my join outputs no data unless I delete the empty input. How do I fix my workflow so the join will output data every month, even if only 1 input has numbers?

 

[griffinwelsh]

@NBPSE You might try adding a union after the join to bring in unjoined records. This will not filter out any records that don't appear in both sources obviously, but hopefully that is not needed in this usecase.

[NBPSE]

Would that double my output total on months where I do have 2 inputs and the join is working as expected? 

[apathetichell]

@griffinwelsh the problem in the users case is that in situations where there is no data in the right datastream - there are no column headers. join will work with no records - but the column headers must be present. union will not fix this.

This is an architect failure - not a join failure. You are saying join a to b - but b might not exist. Ever tried a join on a table which doesn't exist in any sql db?  my fix would be create a dummy datastream just with column headers. use union to add it to my right datastream before my join. continue with my join. union my join and my left outer after join.

[NBPSE]

What would you do to create a dummy datastream? What tools would you use to architect this successfully?

[NBPSE]

Thank you for the suggestion. I used a Data Cleansing tool to replace Null values with zero (Numeric Fields). This allowed the columns to carry forward and the join to work whether one or both inputs have data.

[apathetichell]

I use text input/dynamic rename/union - but sounds like you solved your issue.