Github user mattyb149 commented on the issue:
https://github.com/apache/nifi/pull/1589
Another thing to consider here (although it would be a fair amount of
refactor) is to repurpose this processor as AttributesToRecord. It doesn't add
efficiency as a "record-aware" processor as you'll only be producing one record
from one set of attributes (i.e. per-flowfile), but if you leverage a
RecordWriter you instantly have an AttributesToAnythingThatHasARecordWriter
processor :) So it would supplant AttributesToJSON (and would support
converting types which AttributesToJSON does not do IIRC) and provide CSV
output (your original goal) as well as Avro, FreeFormText, etc. What do you
think?
---
