Subclassing Django’s runserver causes command to be run twice

This week I was creating my own version of the Django management command runserver. Like the built-in, I wanted to run a lightweight development only web server for my Django app. In addition, I was looking to run a syncdb, load some initial data and start up the server with that. The goal was to use an in-memory sqlite3 database, and use this command as a way to spin up an interactive test instance.

This is easily achieved by using call_command to call syncdb and runserver in sequence. The in-memory sqlite3 config I would leave to a Django settings file. But I noticed that syncdb seemed to be getting called twice, leading to double the start up time.

Rooting around in the Django bug database, I found that this was a known issue with the default auto-reloading functionality of runserver. The solution turned out to be a simple as disabling that functionality by passing an option to the native runserver when you use call_command.

import os
from django.core.management import call_command
from django.core.management.base import BaseCommand
from django.core.management.commands.runserver import BaseRunserverCommand


class Command(BaseCommand):
    ''' Runs the built-in Django runserver with initial schema and fixtures
    Note: cannot use auto-reloading because that casues resetdb to be called
    twice.
    '''

    # remove this option from the --help for this command
    option_list = (o for o in BaseRunserverCommand.option_list if o.dest != 'use_reloader')
    help = 'Starts a lightweight web server testing.'
    args = '[optional port number, or ipaddr:port] appname appname'

    def handle(self, addrport='', *args, **options):
        call_command('syncdb', *args, **options)
        # BUG: runserver gets called twice if --noreload is False
        # https://code.djangoproject.com/ticket/8085
        options['use_reloader'] = False
        call_command('runserver', addrport, *args, **options)

Chase Seibert