Opened 3 weeks ago

Last modified 13 days ago

#28752 new Bug

django.setup() should not be runnable multiple times

Reported by: pascal chambon Owned by: nobody
Component: Core (Other) Version: 1.11
Severity: Normal Keywords:
Cc: Aymeric Augustin Triage Stage: Unreviewed
Has patch: no Needs documentation: no
Needs tests: no Patch needs improvement: no
Easy pickings: no UI/UX: no

Description (last modified by Tim Graham)

I've been bitten numerous times by the impredictable behaviour of django when django.setup() was called numerous times.

In the old days I had exceptions, now it's mainly subtle breakages of logging configuration.

I couldn't find, in the issue tracker or the dev mailing list statements about this subject, others than request from other users encountering the

For example #26152 concerned script+importable modules.

The latest case in date for me is pytest-django having troubles with these multiple setup() calls : , due to multiple fixtures attempting this auto-setup.

Would it be OK to make django.setup() idempotent, or even expose an "is_ready" flag for easier introspection ?

-- here are some updates, comments get rejected as spam --

Calling django.setup() multiple times is useless, BUT it can happen in lots of cases, that's why imho this case should be handled by the framework to avoid nasty side effects.

These "duplicate calls" often involve the collision between commands, tests, custom scripts, and external launchers like pytest-django. Plus maybe some corner cases when unittest-style TestCases and pytest-style test functions are mixed in the same project.

Users have to do a real gym to call setup() "at some moment" in all these use cases, yet try to prevent multiple calls of this initialization step (like the if__name__ == "main"' protection). So far my only way out was often to check for (not really undocumented) states of the framework before calling setup().

Change History (13)

comment:1 Changed 3 weeks ago by Tim Graham

Resolution: duplicate
Status: newclosed
Summary: Django.setup() should be idempotentdjango.setup() should be idempotent

I believe this is addressed in Django 2.0 by #27176. If not, please reopen with a minimal project that reproduces the problem you're encountering.

comment:2 Changed 3 weeks ago by pascal chambon

Description: modified (diff)
Resolution: duplicate
Status: closednew

comment:3 Changed 3 weeks ago by Tim Graham

Thanks. Can you explain the use case for calling django.setup() multiple times?

comment:4 Changed 3 weeks ago by pascal chambon

Description: modified (diff)

comment:5 Changed 3 weeks ago by pascal chambon

Description: modified (diff)

comment:6 Changed 3 weeks ago by pascal chambon

Description: modified (diff)

comment:7 Changed 3 weeks ago by pascal chambon

My comments get marked as spam, I updated the description above.

comment:8 Changed 3 weeks ago by Tim Graham

Cc: Aymeric Augustin added
Description: modified (diff)

Aymeric, any input?

comment:9 Changed 3 weeks ago by Aymeric Augustin

django.setup() is already idempotent. However it isn't reentrant — I believe that's what you're actually asking for — and it cannot be made reentrant without breaking its invariants. See #27176 for details.

I don't think we should make changes to Django for accomodating pytest code that does django.setup = lambda: None.

comment:10 Changed 3 weeks ago by pascal chambon

Summary: django.setup() should be idempotentdjango.setup() should not be runnable multiple times

comment:11 Changed 3 weeks ago by pascal chambon

OK I think my vocabulary was wrong, it's not (really) an idempotence problem, since django.setup() does more or less the same things on both calls (just skipping apps population phase on the second).

It's not a reentrancy problem, i.e not a problem with multiple threads (or signal interrupts) entering django.setup() concurrently.

It's really just a problem of "multiple successive calls of django.setup()", which are doing silent errors and weird modifications, simply because only the first call of django.setup() makes sense.

Raising an exception on subsequent calls would be a possibility, but it would be a useless hassle, since users just want is to ensure that django was initialized at some point.
That's why I propose we just do a no-op on subsequent calls to django.setup().

(By teh way I don't understand your statement about the "django.setup = lambda: None" snippet - it was just a quick and dirty hack to prevent the multiple runs of django.setup(), which broke the LOGGING config)

comment:12 Changed 2 weeks ago by Tim Graham

Description: modified (diff)

I don't know. Does that change risk breaking working code where multiple calls to django.setup() has an intended effect?

comment:13 Changed 13 days ago by pascal chambon

Well, I'm usually quite dedicated to retrocompatibility (see django-compat-patcher package), but for once any breakage would be due to a strange misuse of django.setup().

In the code below, we see that django.setup() performs 3 steps :

  • overridding logging configuration with django settings
  • setting the script prefix
  • populating the apps registry

The last 2 steps are now idempotent it seems.

Only overridding logging breaks some setups (eg. with pytest fixtures), and I can't find any use case where it would be a wanted behaviour.
If users want to reset logging several times, they may as well call configure_logging() by themselves.

Until django.setup() is protected against double executions, we'll have weird bugs surfacing each time we add new steps to it (are these idempotent, or reentrant, or runnable multiple times...), so I'd advocate fixing this once and for all.

On "how" to do it, I think a threading lock + a boolean guard would be easy and sufficient, wouldn't they ? With a system to raise an error if django.setup() ends up being called multiple times by the same thread (which often smells like missing "if name == 'main'" conditions in imported scripts).

configure_logging(settings.LOGGING_CONFIG, settings.LOGGING)

if set_prefix:
            '/' if settings.FORCE_SCRIPT_NAME is None else settings.FORCE_SCRIPT_NAME

Note: See TracTickets for help on using tickets.
Back to Top