runserver does not exit when checks fail

[Ben Davis]

Normally I'd think if process of any kind has fatal errors, it should exit with a non-zero status code. If checks result in messages >= ERROR, runserver will just hang there until you press Ctrl+C. This is due to the fact that checks are performed in inner_run() as opposed to outer_run(). Is there a reason why it's done this way, or is this a legitimate bug?

[Russell Keith-Magee]

The check is done on the inner_run() because it's part of the reload mechanism; if you save a file, and an error is found, the inner loop stops, but the outer loop should attempt to reload when you save the file again. So - it's not a bug that the check is done in inner_run().

The fact that runserver hangs *would* be a bug. If you save the affected file and fix the problem, runserver *should* reload. However, I haven't been able to reproduce the problem you describe. I tried creating a sample project with no errors on Django master (9a4bf72f), then I created a duplicate reverse accessor (by duplicating a foreign key), causing a fields.E304 error. When I removed the foreign key, and re-saved the models.py file, runserver restarted as expected.

Closing worksforme; if you can provide details of a specific error where you're seeing this problem, plus details of the Django version etc where you're seeing this, please reopen.