Something in the function tutorial confused me.

Magnus Lycka lycka at carmen.se
Thu Aug 9 16:49:09 CEST 2007


Lee Fleming wrote:
> Hello,
> I have a simple question. Say you have the following function:
> 
> def f(x, y = []):
...

> But this, the code that "fixes" the list accumulation confounds me:
> def  f(x, y=None):
>     if y is None: y = []
...

> In other words, what's going on here? How is it that y accumulates
> argument values between function calls in the first function, but
> doesn't in the second one? 

I think the important thing to understand here is the
distinction between names/variables and objects/values
in Python.

While you could interpret C code like this...

void f() {
     ...
     int i = 5;
     ...
     int j = i;
     ...
}

... as "create a place in the namespace of the f function
where you can fit an integer value, and put the value 5
there. Later, create another place in the namespace of f
which is also big enough for an integer. Copy the contents
of the location named 'i', to the location named 'j'."

You would instead interpret this similar Python code...

def f():
     ...
     i = 5
     ...
     j = i
     ...

... as "create an integer object with the value 5. Then
define a name/tag/variable in the namespace of function
f which refers to the integer object with the value 5.
Later, make a new name/tag/variable in the namespace of
f which refers to the same object (happens to be an
integer with the value 5) as i refers to."

The semantics is very different.

If you understand this, Python will seem much less magical,
and you will never ask meaningless questions as whether
Python uses call by reference or call by value.

It's all a matter of understanding that all the juicy bits
in the Python data model is in the actual values or objects.
That's the stuff with type safety, a location in memory,
qualities such as mutability etc. A "variable" is basically
just a reference to an arbitrary object in a particular
namespace. Assignments semantics is not about copying
data as in C, and it's nothing arbitrarily defined in
classes as in C++. It's all about deciding which object
a name refers to.




More information about the Python-list mailing list