A lot of times I add id columns to incoming CSVs that don't already have one, to make various other tasks simpler later. But when the upstream file changes, and I re-generate a new version of my copy, and then diff that, I can end up with a lot of noise in the case where, say, one line has been moved, and everything after than gets renumbered.

It would be great if there was an option, similar to --uuid, but where the result was more deterministic (perhaps by generating a SHA of the existing data, or something along those lines?)

Or is there some way to get there with qsv apply dynfmt that I can't think of atm (other than explicitly concatenating all the other fields together)?