> Agreed that this could use a good overhaul to make the --help output > less confusing. I'm not sure if this is caused by limitations of the > arguments parsing library used (currently (srfi srfi-37)) or some other > reason.
If we're doing an overhaul of the command line arguments, i would love to see a more "scheme first" approach. At the moment, functionality such as "guix shell" is very difficult to access directly from scheme, and i've usually had to fall back to calling it like (system "guix shell" (concat strings made from s-exps and all)) Ideally, in my humble opinion, the command line command should be a thin wrapper around a scheme-first API.
