On Aug 6, 2015 5:13 PM, "Neal Gompa" <<a href="mailto:ngompa13@gmail.com">ngompa13@gmail.com</a>> wrote: > > On Thu, Aug 6, 2015 at 10:35 AM, Dan Williams <<a href="mailto:dcbw@redhat.com">dcbw@redhat.com</a>> wrote: >> >> On Thu, 2015-08-06 at 08:57 +0200, Aleksander Morgado wrote: >> > On Thu, Aug 6, 2015 at 3:00 AM, Markus Gothe <<a href="mailto:nietzsche@lysator.liu.se">nietzsche@lysator.liu.se</a>> wrote: >> > > Please tell me if I am wrong… But this quite a design issue which we need to come down to consensus… Right? >> > >> > >> > Well, it's not like we're changing the behavior of the kernel driver. >> > This is just a small design issue in mbimcli :) >> > >> > What's your opinion on this? Do you prefer the "=" separated key/value >> > pairs, or just the prefixed "N:" with additional optional arguments >> > for things like ip-type? >> >> I personally prefer = separated key/value pairs, keeping compat with the >> old unkeyed format. So I'll try to do that, but might not happen this >> week. >> >> Dan >> > > If I could chime in for a moment: > > I favor the key/value pair format, as it makes it easier to figure out how to lay out configuration information. Heck, it makes it possible for me to easily structure a program that would take in JSON data representing the configuration and map it straight to mbimcli so that it can be processed. It would be also awesome if some kind of key/value output format could be available as an option for parsing the output that comes back after mbimcli successfully connects. The human-readable output today that shows up in git master is great for humans, but it's a bit tricky to parse for scripting. > > If there's a key/value input and key/value output mechanism, that would make what I'm working on much easier. > Yes, totally agree regarding the output thing. Something like an additional --script option or something that would generate key/value pairs for the outputs of all commands. That would be a truly nice improvement and actually quite trivial to implement.