Why doesn't coverage.py properly measure Flask's runserver command?

Question

I'm trying to figure out which lines of a Flask application are being run. I start Flask like this:

coverage run manage.py runserver

Output looks like this:

 * Running on http://127.0.0.1:5000/
 * Restarting with reloader

manage.py looks like this:

#!/usr/bin/env python
from flask.ext.script import Manager

from my_flask_app import app

manager = Manager(app)

if __name__ == '__main__':
    manager.run()

I then access various parts of the application via HTTP.

When I look at the coverage HTML report, it says only the method definitions are covered, not the actual bodies of the methods.

I suspect it's because the methods are being executed by a subprocess which is not covered by coverage.py.

Any ideas?

Answer 1

So it turns out that the problem is related to the 'reloader' message above. The coverage report is correct when I start Flask like this instead:

coverage run manage.py runserver -R

Output then only contains this:

* Running on http://127.0.0.1:5000/

This way it doesn't start up the server in a separate process, and coverage works great.

I found this solution thanks to this related Django question:

Why doesn't coverage.py properly measure Django's runserver command?