Our scripting system has a bunch of code in the entry_point(), that isn't currently tested by test_wrappers, which will only run the <script>.main(*args) function.
This can mean that bugs slip through if the code in that section is not trivial. There are two alternatives I suppose:
- Rework all of the scripts so that entry_point only contains the code for parsing the command-line args
- Add an additional test for each script, which uses
os.system to call the script from the command line.