Cool that you can download this as a standalone binary for OSX/linux/windows! Nice work

Sometimes we work with really large CSV, anyway to make it bail on first error?

I haven't done any performance testing on large data sets yet, but in general I've been super happy with the Go's CSV library, both in terms of performance and ease-of-use.

If the speed ends up being a problem, it should be pretty easy to add a "--bail" flag or something.

In your experience is it CPU bound or disk? Maybe you can split files and use go routines etc?