Best practice for declaring an empty variable

Question

I know Python is a dynamic language and declaring variables before-hand is generally not necessary, but in the case where I need to create a variable and then give it a value later, both of the following methods work:

some_variable = '' 
some_variable = str()

Is there a difference between both of these and which is considered best practice?

Example:

some_variable = str() 

for number in range(10): 
    some_variable += str(number)
print(some_variable)

Works for both some_variable = '' and some_variable = str()

By best practice I do not mean "which is best coding style/most readable" but rather factors such as memory consumption, speed and which on the whole is more reliable for general use.

Answer 1

If you want to initialize a variable to an empty value, using an empty value literal will be slightly faster, since it avoids having to do a name lookup.

That is, if you write:

some_variable = str()

Python first needs to make sure you haven't written something silly like

str = int

in some visible scope first, so it has to do a recursive name lookup.

But if you write:

some_variable = ''

Then there's no way that '' will ever be anything but a str. (Similar principles apply for list and tuple literals: prefer [] and () over list() and tuple().)

More generally, though: initializing a variable to an empty value is generally considered a code smell in Python. For things like an empty list, consider using a generator expression (... for ... in ...) or a generator function (using yield) instead -- in most cases you can avoid setting up empty values at all, and that's generally considered a more Pythonic style.

Answer 2

I know Python is a dynamic language and declaring variables before-hand is generally not necessary

What? First of all how is declaring a variable before hand related to being dynamic or not? Secondly how can you even imagine any language that forbits declaring a variable "before-hand" (whatever that means) to work? You need to declare some variable directly at some point.

Is there a difference between both of these and which is considered best practice?

Yes, there are differences. Consider this:

def str():
   return 1

x = str()

Python does not protect you from variable shadowing. But the literal empty string '' cannot be shadowed. That's a difference on the functional level.

There is also a difference on the performance level: x = '' is slightly faster then x = str() due to being translated to a different bytecode. After all x = '' does not require a function call (which cannot be optimized again due to variable shadowing).

And finally esthethics: x = '' is probably more readable then x = str(). Again since you need to be sure what str() actually is in the context.

So all in all: go with x = ''.