Opened 13 years ago
Closed 11 years ago
#17795 closed Cleanup/optimization (wontfix)
kwargs not passed on by django.views.generic.edit import ProcessFormView
Reported by: | Owned by: | Adrien Lemaire | |
---|---|---|---|
Component: | Generic views | Version: | dev |
Severity: | Normal | Keywords: | |
Cc: | lemaire.adrien@…, bmispelon@…, polmuz | Triage Stage: | Accepted |
Has patch: | yes | Needs documentation: | yes |
Needs tests: | no | Patch needs improvement: | no |
Easy pickings: | yes | UI/UX: | no |
Description (last modified by )
Although these do get attached to self.kwargs for use within the class, this causes problems for decorators of the get_context_data method.
kwargs which you may expect to be available for use by decorators you create are unset by this mixin.
NB: In my case this was related to testing user object permissions - hence the need to have the context data
Patching as below fixes it:
A mixin that processes a form on POST. """ def get(self, request, *args, **kwargs): form_class = self.get_form_class() form = self.get_form(form_class) -- return self.render_to_response(self.get_context_data(form=form)) ++ kwargs['form'] = form ++ return self.render_to_response(self.get_context_data(**kwargs))
Attachments (3)
Change History (24)
comment:1 by , 13 years ago
comment:3 by , 13 years ago
Description: | modified (diff) |
---|
comment:4 by , 13 years ago
Triage Stage: | Unreviewed → Accepted |
---|
comment:5 by , 13 years ago
Cc: | added |
---|---|
Owner: | changed from | to
comment:6 by , 13 years ago
Has patch: | set |
---|---|
Needs tests: | set |
by , 13 years ago
Attachment: | t17795.diff added |
---|
Add kwargs to context in all generic base classes get() + test
comment:7 by , 13 years ago
Needs tests: | unset |
---|---|
Patch needs improvement: | set |
@ed.crewe , I couldn't reproduce your problem, I don't understand how you're testing your user object permissions. I always do it through get_object or dispatch, and didn't get such problem.
I wrote a test that does basically the same thing, by adding a kwarg in an overrided get() instead of get_context_data, and insure that this new kwarg isn't lost in ProcessFormView.get(). I don't like my test, so if you could give me your use case, I'll replace the test by a proper one.
I'm wondering if I should do the same with post ? Again, I need to understand the use case before doing so.
by , 13 years ago
Attachment: | example_code.py added |
---|
example decorator needing objects and monkeypatches to provide this
comment:8 by , 13 years ago
Hi,
I have attached a file with the monkey patches I am using of class views - and an example decorator that demonstrates the issue.
So why it is desirable for get_context_data to include the views object(s) in the kwargs it is passed - rather than just getting them from the view attributes, ie. self.object - so that standard decorators for the context data method can be used which need access to a views object(s) or possibly other attributes.
In this case an object permissions test decorator.
Thanks,
Ed
comment:9 by , 13 years ago
Patch needs improvement: | unset |
---|
I think we're good now. I wrote responsible tests thanks to @acdha idea of using RequestFactory for that case.
I realize that there is no test for BaseDateListView, so I omitted that one, let me know if I should write tests for the archive views.
comment:10 by , 13 years ago
Patch needs improvement: | set |
---|
The patch looks good. If you could add tests for BaseDateListView
that would be great. Also some minor comments:
- Verify the actual value of the 'foo' context variables instead of just verifying their existence.
- Instead of updating
kwargs
(e.g.:kwargs['form'] = form
) before passing it toself.get_context_data()
, use named arguments, i.e.:self.get_context_data(form=form, **kwargs)
, mainly for consistency with the rest of the codebase.
by , 13 years ago
Attachment: | t17795.2.diff added |
---|
kwargs for get_context_data() result, and factory tests
comment:11 by , 13 years ago
Patch needs improvement: | unset |
---|
Done. I just needed to write a test for one of DateBaseListView
subclasses, I wrote one for ArchiveIndexViewTests.
actual value of 'foo' is verified.
BaseDetailView.get() now consistent with the rest of the code.
comment:13 by , 12 years ago
Cc: | added |
---|
For reference, commit f04bb6d798b07aa5e7c1d99d700fa6ddc7d39e62 added this behavior to TemplateView
.
I'm wondering if it'd be a good idea to just add this behavior in ContextMixin
instead of re-implementing it in each view.
comment:14 by , 12 years ago
Patch needs improvement: | set |
---|---|
Triage Stage: | Ready for checkin → Accepted |
I'm not sure if this requires any changes per the last comment, but the patch no longer applies cleanly.
comment:15 by , 11 years ago
Cc: | added |
---|---|
Patch needs improvement: | unset |
Triage Stage: | Accepted → Ready for checkin |
Version: | 1.4-beta-1 → master |
Hi, I created a PR that merges cleanly using the patch provided. I think it's ready for checking again given that adding the behavior in ContextMixin doesn't seem to be straightforward.
comment:16 by , 11 years ago
Needs documentation: | set |
---|---|
Triage Stage: | Ready for checkin → Accepted |
This behavior change needs a mention in the release notes and probably the docs themselves.
comment:17 by , 11 years ago
I want to -1 this change. URL keyword arguments are not context data. IMO the behaviour of TemplateView is wrong here, and has bitten me before where I've had to change my urlconf.
For a concrete exple of why this is bad, consider the situation where I have a URL pointing to a form view which renders different forms depending on the url. Then I call that URL kwarg 'form' and this code breaks.
The correct place to add things to the context is in get context data, not by arbitrarily adding all URL kwargs to the context in all views.
This patch would break existing running code for me.
As an aside, note that view is put into the context now, so if you really want to you can access the URL kwargs as view.kwargs
comment:19 by , 11 years ago
@mjtamlyn: +1, it bit me too. It's a problem with TemplateView which if I remember well, did that for backward compat with the old direct_to_template
.
It'd be great to deprecate that...
comment:20 by , 11 years ago
That's what #19878 is about. the details of the deprecation are tricky.
comment:21 by , 11 years ago
Resolution: | → wontfix |
---|---|
Status: | new → closed |
Marking as won't fix given the backwards compatibility concerns.
Same issue in
dates.BaseDateListView.get
list.BaseListView.get
detail.BaseDetailView.get