Is this unpythonic?

[Frank Millman]

"Frank Millman" <frank at chagford.com> wrote in message 
news:mik7j6$59n$1 at ger.gmane.org...
>
> "Steven D'Aprano" <steve+comp.lang.python at pearwood.info> wrote in message 
> news:554cd119$0$12977$c3e8da3$5496439d at news.astraweb.com...
>> On Fri, 8 May 2015 08:53 pm, Frank Millman wrote:
>>
>>>> Does z have to be a list? Could you use an empty tuple instead?
>>>>
>>>> def x(y, z=()): ...
>>>>
>>>
>>> That was Chris' suggestion as well (thanks Chris).
>>>
>>> The idea appealed to me, but then I found a situation where I pass in a
>>> dictionary instead of a list, so that would not work.
>>
>>
>> Why wouldn't it work? If it worked with an empty list, it will probably 
>> work
>> with an empty tuple instead.
>>
>
> Sorry, I should have been more explicit. In the case of a dictionary, I 
> used 'def x(y, z={}'
>
> I have not checked, but I assume that as dictionaries are mutable, this 
> suffers from the same drawback as a default list.
>
> Unlike a list, it cannot be replaced by an empty tuple without changing 
> the body of the function.
>
> Dave's suggestion would have worked here -
>
>    EMPTY_LIST = []
>    EMPTY_DICT = {}
>
> But as I have decided to use the None trick, I use it for a default 
> dictionary as well.
>

Cough, cough, I really should have given that a moment's thought before 
posting.

It just dawned on me that a dictionary *can* be replaced by an empty tuple 
without changing the body of the function.

There are two operations I might perform on the dictionary -

1. iterate over the keys and retrieve the values

2: use 'in' to test if a given string exists as a key

Both of these operations will work on a tuple and give the desired result, 
so it is a very valid workaround.

More testing needed ...

Frank