Currently, junk dimensions populate DW_SourceCode with the values from the source table that contributes the rows into the junk table.
For a junk dimension that is only populated from one source (fact) table, this definitely makes sense.
It probably also makes sense in _some_ scenarios when multiple fact tables are populating a junk table. It could be interesting or important to know which fact table led to a particular junk row being added.
However - in some junk dimension usage, which source fact table the row comes from is unimportant, or even actively misleading - and one fact table has the same combination of junk attributes 2+ times, from different sources, then it causes duplicates in the junk RAW table, and a PK violation error to be shown in the 'Errors' tab/report.
In our usage at least, this error row is misleading and distracting.
So, I think it would be great if junk dimensions came with a configuration option to ignore the DW_SourceCode when populating the junk table.
I still think the Junk table should have a DW_SourceCode column in it, but if this option is set, then the DW_SourceCode should be populated with a static value like 'Junk Dim' or similar. The JunkFill procedure would not include DW_SourceCode in the select column lists.