In all versions of CPython right now, the following works.

>>> dict({1:2}, **{3:4})
{1: 2, 3: 4}

Other Python implementations raise TypeError for this;  CPython should probably do the same, beginning with deprecating this behaviour in Python 3.2 and removing it in 3.3.

From a python-dev posting[1] by Alex Gaynor:

"""
I ran into the follow behavior while making sure Django works
correctly on PyPy.  The following behavior was observed in all tested
versions of CPython (2.5, 3.1):

>>> def f(**kwargs):
...     print(kwargs)
...
>>> kwargs = {1: 3}
>>>
>>> dict({}, **kwargs)
{1: 3}
>>> f(**kwargs)
Traceback (most recent call last):
 File "<stdin>", line 1, in <module>
TypeError: f() keywords must be strings
>>>

This behavior seems pretty strange to me, indeed PyPy gives the
TypeError for both attempts.  I just wanted to confirm that it was in
fact intentional.
"""

Raghuram Devarakonda says (in the same python-dev thread):

"I ran into same issue with Django on Jython yesterday [1] since Jython
too gives TypeError for 'dict({}, **kwargs)'."

Guido, on the suggestion that both the CPython and PyPy behaviour be left as is, and that the behaviour be regarded as implementation defined:

"That is just going to cause some programs to have a portability
surprise. I think one or the other should be fixed. I am fine with declaring dict({}, **{1:3}) illegal, since after all it is abuse of the ** mechanism. We should deprecate it in at least one version though."


[1] http://mail.python.org/pipermail/python-dev/2010-April/099427.html

The issue applies to dict.update as well:

>>> d = {1:2}
>>> d.update({3:4}, **{5:6})
>>> d
{1: 2, 3: 4, 5: 6}

Update:  IronPython also produces a TypeError here (thanks Michael Foord and Dino Viehland).

I don't think this is worth the performance hit or worth the deprecate/remove exercise.  I agree with Guido that we just call this an undefined, implementation specific behavior.

This behavior has existed for a very long time with no ill-effects.  Marking as low priority.

> I agree with Guido that we just call this an undefined, implementation specific behavior.

Actually I think that was exactly what Guido was *disagreeing* with.  :)

I agree with Mark. Guido's point was that if it's implementation defined, you'll have portability problems: http://mail.python.org/pipermail/python-dev/2010-April/099435.html

Guido seems to be favoring disallowing it, not ignoring it[0].

http://mail.python.org/pipermail/python-dev/2010-April/099435.html

Does this mean that every time someone uses **kwds, that the entire dictionary will need to be scanned for invalid keys?  So every API that accepts and passes through **kwds will be slowed on every time it is called.  Is there any offsetting benefit?

They are already checked for pure python functions.

Here's a patch. It adds a new function, PyArg_ValidateKeywordArguments.

And here's a patch for the correct Python version...

Now with the suggested dictionary speed hack.

Fixed in r80450.