What is the use of 'Propagate Columns' feature in SAS DI studio

[eagles_dare13]

What does this feature do? 'Propagate Columns from Source to Target'? Does it copy all the columns of the source table to the target tabl

[DaveR_SAS]

This topic in the DI Studio user guide covers some of the basics: SAS(R) Data Integration Studio 4.8: User's Guide

[eagles_dare13]

Thanks for the link but if you don't mind, can I ask:

"You have added columns and you need to determine the scope of these additions"

What is meant by this? I have a source table and I have a target table. So where am I adding columns? And what is meant by 'scope' here?

I understand the concept of mapping the columns between source and target but not able to understand the 'propagate ' feature.

[LinusH]

It means, what do you want to do with the columns that you added (can be applied to columns added to any table in any job).

[Quentin · Accepted solution]

Hi,

I'll take a stab.  But I don't actually use DI studio transformations much (usually do more with user-written code).

As I understand it, if you have a traditional DIS job, you often have a number of transformations connected in a process flow.  So maybe first transformation pulls some data from a database, second transformation joins that data to some other data, third transformation transposes it and outputs the final dataset produced by the job.

So suppose you have that job, and it's running fine.

Then one day somebody wants a new variable in the output dataset.

You go to the first transormations (data pull), and add the extra column to that.  Then the question is do you want that new column to be carried forward throughout your flow.  If yes, you could propagate it forward throughout the flow.  If you don't propagate it forward, the new variable could be pulled during the the first transformation, but excluded by the second transformation.

Similarly, I think you may be able to change attributes of a variable (format/label/etc) and choose whether or not to propagate those changes.

Hope that sounds reasonable.

[eagles_dare13]

Thanks @Quentin. Your answer cleared my doubts.

[eagles_dare13]

Can you also explain what it means to Propagate Job?

[DaveR_SAS]

To me, this topic. . .

SAS(R) Data Integration Studio 4.8: User's Guide

. . . suggests that Propagate Job is another way of propagating columns in the current job. It sounds similar to Propagate Columns as described in this table. Perhaps Propagate Columns would also handle the case where a job had more than one flow, for example.

[DaveR_SAS]

Darn it. This discussion app does not seem to link to subheadings within a topic. Please see the table under "Manual Propagation" in the topic I linked to.