Flatten & Multiline transform options when creating a DQ dataset

I received Zendesk ticket #99042 from our partner Lucid in regard to the transform options for “Flatten” & “Multiline” while building out a DQ dataset.

I asked in #dq-forum about this but still a bit of confusion on what both of these options do, what data types/sources you can use them on, & best practices on using them.

Eric Gerstner was able to lead me to this jira and from my testing, it seems like flatten just parses out the basic structure of the file while multiline suggests that there are additional key-value pairs within the flattened object?

Can someone confirm that this is true? If so can these additional questions be answered as well:

1. Would there be a reason why you would use multiline without flattening the file?
2. Can this be used on remote file systems, file uploads, & temp files? I don’t believe these work for any JDBC-connected data sources, correct?
3. Is there a limitation and/or recommendation on how many levels will work with these transform options?

Thanks!