Python functions: Pass global variables if only accessing them?

问题

I'm trying to get a sense of best practices in python. If I define a function to access (but not change) a global variable, it's not generally necessary to specify the variable as global in the function. But is it faster to pass the global variable to the function? I ask because I've come across some references to the cost of looking up global variables, but I'm not sure that I understand. For example:

def f1(localList):
    for element in localList:
        if element in globalSet:
            pass #do stuff.

def f2(localList, localSet):
    for element in localList:
        if element in localSet:
            pass #do stuff.

globalList = <arbitrary list>
globalSet  = <arbitrary set>

f1(globalList)
f2(globalList, globalSet)

is f2 generally considered to be the faster/better/more "pythonic" approach compared to f1?

回答1:

If globalList and globalSet data is tightly bound to the functions i'd use a class and define those lists as class propierties and those functions as staticmethods (or instance methods, whatever i need)

回答2:

Global variables are generally considered bad practice in programming in general. Basically because any code that can see the variable can change it. Especially in multithreaded programs this can be a source of bugs.

You should think of Python variables as names for, or references to objects. So passing a variable to a function isn't costly.

回答3:

Using Globals is always a bad design choice. In any case you can always come up with a solution which would be more Pythonic by not using Globals.

• If you have to share lot of variables between functions, you can reconsider an OO approach.
• If you have to pass many parameter's to a functions, you can opt for varargs or kargs.

Remember, parameters are passed by reference so there is no efficiency lost.

----Special cases aren't special enough to break the rules.

来源：https://stackoverflow.com/questions/10234901/python-functions-pass-global-variables-if-only-accessing-them