I am trying to process a large CSV file using a map/reduce script. In my getInputData() step, I am returning a file.load() of my CSV file. Due to new functionality in 2017.1, each "map()" step will then be called on each row of the CSV file passed from getInputData() My question to this group: How can I also-pass the first row (column headers) to each map() step? I've tried using n/cache, however, that appears to nullify itself after about an hour or two of runtime. Aside from silliness of writing a file in the file cabinet (then reading that from every map() step) is there a better way to pass "metadata" from getInputData() down to each map() method?