modelform_factory, modelformset_factory, inlineformset_factory formfield_callback lambda function missing **kwargs

[Harro]

When using the inlineformset_factory with a custom modelform class which has the widgets dict defined in the meta class for one of the fields I ran into the following error.

<lambda>() got an unexpected keyword argument 'widget'

Some debugging later I discovered the formfield_callback definition for the inlineformset_factory is missing the kwargs argument.

[Harro]

Bug in inlineformset_factory patch

[Russell Keith-Magee]

Not critical for 1.2

[devinj]

Updated patch, also fixes modelform_factory and modelformset_factory. Adds tests. Removes encoding problems of previous attempt.

[Harro]

I've tested devinj's patch, the regression test fails without the fix, it succeeds with the fix.

From my own code I can confirm that the patch works.

[Simone]

for the current trunk 13307

[Simone]

Ok is not critical for 1.2,

but which is the best solution to have a workaround?
We cannot use the Meta widgets = {} in the form...
Ok, we can override the init method and override the widget.

[Russell Keith-Magee]

I'll probably end up fixing this as part of #13633, since there will be some overlap (especially when it comes to adding tests)

[Russell Keith-Magee]

#13633 has been closed wontfix, so this problem needs to be solved on its own. @kenth provided another patch on that ticket.

[jk]

I get an error when I try to apply the patch

patch --dry-run -p1 < ~/Downloads/ticket13095_r12761.diff 
patching file forms/models.py
Hunk #1 FAILED at 797.
1 out of 1 hunk FAILED -- saving rejects to file forms/models.py.rej

Any ideas?

[jk]

Replying to jk:

I get an error when I try to apply the patch

patch --dry-run -p1 < ~/Downloads/ticket13095_r12761.diff 
patching file forms/models.py
Hunk #1 FAILED at 797.
1 out of 1 hunk FAILED -- saving rejects to file forms/models.py.rej

Any ideas?

My apologies.
I tried applying the patch "ticket13095_r13307.diff" and it worked. It did complain about the missing test files, which I skipped.

[vung]

Use None as default value for formfield_callback (patch from #13633+tests)

[vung]

Having None as default for formfield_callback seems better to me, so I added some tests to the patch provided by @kenth in #13633.

The patch is a bit different, though. If formfield_callback is None, then formfield() is called directly instead of creating a lambda that calls it.

Note that not all tests fail before the patch; some of them are there to ensure that changing the defaults to None doesn't change current behaviour.

[Carl Meyer]

Verified that the latest patch (from vung) applies cleanly, fixes the reported problem, improves code maintainability (the same non-trivial default should not be repeated several different places, or we're asking for this bug to come back for a repeat visit), and adds useful test coverage (I verified that messing with any of the formfield_callback defaults now causes at least one test failure).

Entire test suite passes with the patch (tested only on sqlite, but it's not an ORM patch).

Marking ready-for-checkin. (Feel free to slap me if this is overstepping things).

[jbronn]

(In [13730]) Fixed #13095 -- formfield_callback keyword argument is now more sane and works with widgets defined in ModelForm.Meta.widgets. Thanks, hvdklauw for bug report, vung for initial patch, and carljm for review.

[jbronn]

(In [13731]) [1.2.X] Fixed #13095 -- formfield_callback keyword argument is now more sane and works with widgets defined in ModelForm.Meta.widgets. Thanks, hvdklauw for bug report, vung for initial patch, and carljm for review. Backport of r13730 from trunk.

[Simon Litchfield]

See related tickets #14572 and #14574