Opened 17 years ago
Closed 10 years ago
#7836 closed Bug (fixed)
django.test.client._session should check for session middleware instead
| Reported by: | Trevor Caira | Owned by: | nobody | 
|---|---|---|---|
| Component: | Testing framework | Version: | dev | 
| Severity: | Normal | Keywords: | |
| Cc: | charette.s@…, vlastimil@…, unai@… | Triage Stage: | Accepted | 
| Has patch: | no | Needs documentation: | no | 
| Needs tests: | no | Patch needs improvement: | no | 
| Easy pickings: | no | UI/UX: | no | 
Description
The django test client class allows you to use self.client.session only if you have 'django.contrib.sessions' in your INSTALLED_APPS setting. That is, it assumes you are using sessions if and only if you have the sessions app installed.
This assumption breaks down when you are using the sessions middleware independently of the sessions app, for instance if you are using a cache backend for sessions.
I propose that django.test.client._session check for the sessions middleware class in MIDDLEWARE_CLASSES instead of the sessions app in INSTALLED_APPS to make it easier for people using the session middleware independently of the sessions app to use self.client.sesssion in tests.
Change History (15)
comment:1 by , 17 years ago
| Component: | Uncategorized → Unit test system | 
|---|
comment:2 by , 17 years ago
| Triage Stage: | Unreviewed → Design decision needed | 
|---|
comment:3 by , 15 years ago
| Triage Stage: | Design decision needed → Accepted | 
|---|
comment:4 by , 15 years ago
| Severity: | → Normal | 
|---|---|
| Type: | → Bug | 
comment:7 by , 14 years ago
| Easy pickings: | unset | 
|---|---|
| Resolution: | fixed | 
| Status: | closed → reopened | 
| UI/UX: | unset | 
comment:8 by , 13 years ago
| Cc: | added | 
|---|
comment:9 by , 13 years ago
| Cc: | added | 
|---|
comment:10 by , 13 years ago
| Status: | reopened → new | 
|---|
follow-up: 13 comment:11 by , 12 years ago
I'm probably missing something, where's the need of checking if some kind of session middleware exists? If not present, wouldn't {} be returned anyways?
BTW, shouldn't it be better if the engine module import happen outside _session and instead of checking for if coockie if cookie is not None is checked?
comment:12 by , 12 years ago
| Cc: | added | 
|---|
comment:13 by , 12 years ago
Replying to unaizalakain:
I'm probably missing something, where's the need of checking if some kind of session middleware exists? If not present, wouldn't
{}be returned anyways?
In application, the attribute request.session is created by SessionMiddleware. If missing you will get AttributeError.
In tests, you get either empty dictionary or SessionStore.
The problem here is no error is indicated by client when you don't have sessions installed, but client.login does not do a valid login. This is problem which is not quite easy to find.
BTW, shouldn't it be better if the engine module import happen outside
_sessionand instead of checking forif coockieif cookie is not Noneis checked?
It wouldn't. You can change settings for the test and you want the test to use yours.
comment:14 by , 12 years ago
What if _sesion arises an exception if no subclass of the session middleware is found?
comment:15 by , 10 years ago
| Resolution: | → fixed | 
|---|---|
| Status: | new → closed | 
I think that the recent commit 21dd98a3 fixing #25699 is also fixing this issue.
This looks like a reasonable request.