I like QueryRecord SQL-based transformation capabilities in Nifi via its Calcite implementation. But it seems in competition with the power and ease of use of Nifi Expression language in UpdateRecord. I am looking for perspective on how other people are approaching this with respect to 1) performance on large flowfiles, and 2) robustness/practicality of committing to one approach vs. the other.
So far, I have been unsuccessful in figuring out how to turn on dialect-specific operators in Nifi as mentioned in the Calcite docs, which would tilt the scales in favor of QueryRecord. For now, my approach is to use both processors - QueryRecord followed by UpdateRecord - but I would like to choose one over the other.