I've been running into some of this again with the plots.

A few thoughts:

Regarding parse_args vs. parse_known_args: I agree the sub-scripts (run_scenario, process_results, ...) being called in run_end_to_end.py need to use parse_known_args, but everywhere else (run_end_to_end.py, ...) we should be able to use parse_args instead of parse_known_args.
I think we should use the parent parser construct to set up a parser once for common arguments such as database, scenario_id, scenario and scenario_location (rather than repeating all that code in every sub-script). The viz suite has an example of how this is done.
We can make argument requirements/types/etc. a little more explicit when setting up the ArgumentParser objects by also specifying the type, choices (if applicable), a 'required', and more for each argument. Note that the documentation does not recommend using "--" type keywords for required arguments since the "--" flag is generally assumed to denote optional arguments.
Related to (3), I think we should add the default for scenario_location explicitly when setting up the parser, rather than assigning it in gridpath.common_functions.determine_scenario_directory
We should handle the case of non-existing scenarios or scenario_ids better (right now it propagates to the SQL statement in auxiliary.auxiliary.get_scenario_id_and_name(), where you query the scenario table but don't get any matches so the fetchone()[0] statement doesn't work.

Error-handling for missing or incorrectly specified script arguments #111

Description