You’ve got attributes that you keep track of (username, checkout time, feature name, version, host, etc..). By default, if you invoked say “lmstat –a –csv” it would simply output all attributes with a header describing the attribute names:
JSON and XML are self-documenting. I would imagine if you actually polled people they would prefer JSON because it is more compact, but I definitely think a CSV output would be appealing to many.
It would also be nice to either be able to enumerate only the fields you want via command line or provide an input file that lists the fields you want.
The main goal of this output is to provide a format-invariant, easily machine parsable version of usage which could feed some form of database or analytic engine.
