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

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.