Stack OverflowUsing global variables in a function other than the one that created them
[+1971] [16] user46646
[2009-01-08 05:45:02]
[ python global-variables scope ]
[ ]

If I create a global variable in one function, how can I use that variable in another function?
Do I need to store the global variable in a local variable of the function which needs its access?

[+2786] [2009-01-08 08:39:44] Paul Stephenson [ACCEPTED]

You can use a global variable in other functions by declaring it as global in each function that assigns to it:

globvar = 0

def set_globvar_to_one():
    global globvar    # Needed to modify global copy of globvar
    globvar = 1

def print_globvar():
    print(globvar)     # No need for global declaration to read value of globvar

print_globvar()       # Prints 1

I imagine the reason for it is that, since global variables are so dangerous, Python wants to make sure that you really know that's what you're playing with by explicitly requiring the global keyword.

See other answers if you want to share a global variable across modules.

(10) The same thing applies to PHP to make sure you're aware they're global - Aram Kocharyan
(469) It's extreme exaggeration to refer to globals as "so dangerous." Globals are perfectly fine in every language that has ever existed and ever will exist. They have their place. What you should have said is they can cause issues if you have no clue how to program. - Anthony
(120) I think they are fairly dangerous. However in python "global" variables are actually module-level, which solves a lot of issues. - Fábio Santos
(11) @Anthony, "has ever existed and ever will exist", don't you think that statement is kind of broad?! :-D IIRC some languages don't even have globals. - Prof. Falken
(38) I was using hyperbole, but I think my underlying point is still valid. - Anthony
(4) @Anthony - Please supply one case where it's a good idea to use globals in Python. I'm using it right now because I'm just hacking something together, but as soon as I see it working I'll be finding a proper way of passing the argument to my function (or I'll see it doesn't work and give up on having this feature...) I'm personally surprised it's in Python... most of the time it seems like Python doesn't have such constructs and forces you to do things the smart/safe way the first time. - ArtOfWarfare
(39) @ArtOfWarfare Aren't set_globvar_to_one and print_globvar globals too? Everybody freaks out about global scalars (lists, objects, etc), but not global functions, which are first-class citizens, too. - Hyperboreus
(4) @ArtOfWarfare how about this - you have a variable which holds a large amount of data. you want to use several functions to manipulate the data in this variable and you also want to conserve memory. wouldn't globals be the best way of doing this since multiple large local variables would take up more memory? - mulllhausen
(4) @Hyperboreus: how often do you see the code like global some_function? Most functions once defined are never changed. Moreover, good interfaces try to expose only as many names as necessary. __all__ may contain only a tiny part of all available functions in a module. - J.F. Sebastian
(6) Try reading in a massive dictionary file and passing it around on recursion function calls. See if you run out of stack space, even after doing the hacky sys.setrecursionlimit. Globals have their place. - mdenton8
(2) Also relevant: python's obviously not for real programs, but it's useful to use concurrency in python for even the most menial tasks (a downloader or scraper?) and sometimes globals then become necessary. - mdenton8
(3) @Hypeboreus: functions are global constants and not global variables! - Daniel
(3) @mullhausen: global variables has nothing to do with the amount of memory. Since all variables in python are only references, many local variables can share the same memory - Daniel
(6) @mdenton8: if you have a problem with the recursion limit, global variables don't solve your problem, but add another one. Try to use loops. - Daniel
(80) I disagree that the reason Python requires the global keyword is because globals are dangerous. Rather, it's because the language doesn't require you to explicitly declare variables and automatically assumes that a variable that you assign has function scope unless you tell it otherwise. The global keyword is the means that is provided to tell it otherwise. - Nate C-K
(4) @NateC-K: Good point. But the interesting comparison here is not with something like C, which requires you to declare variables, but with something like JavaScript, which has implicitly-declared variables but defaults to globals. (Global variables aren't inherently dangerous, but accidental global variables are…) - abarnert
(1) @mullhausen & others, global variables are quick and dirty programming in Python and OOP world. The orthodox way is to use model-view-control pattern. This also 'forces' to use getters, setters and '@property'. The end result is the same: you have a global object which is seen by all other objects. The drawback is that you have to create the model object (containing the huge data set) and put all your manipulating functions into control objects and tell each control object where the model object is. After saying this, it is just much faster to add word global in front of the variable. - Juha
@Juha global variables are not always dirty. We can implement global variables through OOP and it has it's merits. The reverse also has it's merits. In my opinion, using global variables is far better than going for multiple inheritance. - thiruvenkadam
In Reactive Programming with JavaScript, I opened by discussing Crockford's good/bad parts and saying in essence that individual details admit legitimate debate, but some attempt to draw a line that is necessary, and I can count on one finger the number of professional JavaScript authors who condescend to call Douglas Crockforf naive for being a central figure, Crockford being silly enough to say that we aren't paid to use all the features; we are paid for a turning in solid work with no obligation to use problematic features that are snares. - JonathanHayward
(1) global in python helps performance, because you search less, and search the right areas first - dynamic languages spend a lot of time looking for variables. global also reduces bugs, because you don't have a typo that uses a variable without knowing. You should be aware of when you're using a global. When I'm programming in C, all my globals start with g_ . - Brian Bulkowski
such pain.... need to be done everywhere! - Bhabani Sankar Mishra
@mdenton8: Passing a large (or small) dictionary as an argument to another function (including recursive call to the same function currently executing) doesn't make a copy of it each time. - martineau
@martineau Yes I perhaps misunderstood that at the time, but my comment still holds true as a pointer to that dictionary will now be on every stack frame of a recursive call. Python seems to create so much information per stack frame that it is still a large overhead. Globals can avoid this overhead and avoid stack overflows of recursion. - mdenton8
@mdenton8: Sorry, I generally disagree. The trivial amount of additional overhead that comes with adding an extra argument is unlikely to make a significant difference one way or the other, even for highly recursive algorithms. Rather indicating the need to use globals, it more likely indicates that the algorithm should modified or even replaced with one that is non-recursive (or at least less so). - martineau
@martineau "generally" doesn't matter. I have written a recursive function that overflowed the stack by passing around extra arguments, that was very difficult to write non-recursively and significantly easier to understand when written recursively. So globals were by far the best option. As python is meant to be expressive rather than performant, an easy to read recursive function was by far the best option, and the only solution to the stack overflow was a global. - mdenton8
@martineau That specific case was of course the reason I was commenting on this question back then. So, "I generally disagree" - generally doesn't matter. "trivial amount of additional overhead" - most overhead on the stack, especially because python adds extra, is non-trivial with python's small stack size. "Rather indicating the need to use globals" - it did. - mdenton8
[+571] [2009-01-08 09:19:55] Jeff Shannon

If I'm understanding your situation correctly, what you're seeing is the result of how Python handles local (function) and global (module) namespaces.

Say you've got a module like this:

myGlobal = 5

def func1():
    myGlobal = 42

def func2():
    print myGlobal


You might expecting this to print 42, but instead it prints 5. As has already been mentioned, if you add a 'global' declaration to func1(), then func2() will print 42.

def func1():
    global myGlobal
    myGlobal = 42

What's going on here is that Python assumes that any name that is assigned to, anywhere within a function, is local to that function unless explicitly told otherwise. If it is only reading from a name, and the name doesn't exist locally, it will try to look up the name in any containing scopes (e.g. the module's global scope).

When you assign 42 to the name myGlobal, therefore, Python creates a local variable that shadows the global variable of the same name. That local goes out of scope and is garbage-collected [1] when func1() returns; meanwhile, func2() can never see anything other than the (unmodified) global name. Note that this namespace decision happens at compile time, not at runtime -- if you were to read the value of myGlobal inside func1() before you assign to it, you'd get an UnboundLocalError, because Python has already decided that it must be a local variable but it has not had any value associated with it yet. But by using the 'global' statement, you tell Python that it should look elsewhere for the name instead of assigning to it locally.

(I believe that this behavior originated largely through an optimization of local namespaces -- without this behavior, Python's VM would need to perform at least three name lookups each time a new name is assigned to inside a function (to ensure that the name didn't already exist at module/builtin level), which would significantly slow down a very common operation.)


(51) Good explanation. I find it convenient to remind myself from time to time that, in python, "assignment is not an operator." (Contrary to, say, C++, which is where I spend most of my programming time, hence the need for frequent reminders.) This leads to some apparently paradoxical results -- if myGlobal were an array -- [5] -- instead of a scalar, and func1 did myGlobal.append(42), then func2 would print [5, 42] even without a "global" declaration. - c-urchin
(3) Your answers is like meditation. I want to be still and meditate on this answer. This is so good. - NullException
Here is what is most surprising to me: if you modify func2() to print myGlobal then modify it then print it again, then you get an error as well. - Matyas
(1) Yep, good answer. If you understand how the language is interpreted (read books about language parsing) it explains a lot why the global keyword is needed. - tiktak
Great comment about the optimization of local namespaces!! - A.Wan
You mentioned that the namespace decision happens at compile time, I don't think it is true. from what I learn python's compilation only checks for syntax error, not name error try this example def A(): x+=1, if you don't run it, it will not give UnboundLocalError, please verify thank you - watashiSHUN
[+147] [2009-01-08 05:59:04] gimel

You may want to explore the notion of namespaces [1]. In Python, the module [2] is the natural place for global data:

Each module has its own private symbol table, which is used as the global symbol table by all functions defined in the module. Thus, the author of a module can use global variables in the module without worrying about accidental clashes with a user’s global variables. On the other hand, if you know what you are doing you can touch a module’s global variables with the same notation used to refer to its functions, modname.itemname.

A specific use of global-in-a-module is described here - how-do-i-share-global-variables-across-modules [3]:

The canonical way to share information across modules within a single program is to create a special configuration module (often called config or cfg). Just import the configuration module in all modules of your application; the module then becomes available as a global name. Because there is only one instance of each module, any changes made to the module object get reflected everywhere. For example:


x = 0   # Default value of the 'x' configuration setting


import config
config.x = 1


import config
import mod
print config.x

(1) It seems to me like it would be cleaner to express these configuration variables using the ConfigParser class. In fact, that's what I'm trying to do at the moment, and I can't seem to figure it out. - g33kz0r
(5) ConfigParser is not related to sharing global variables between functions; You will still need a "config" instance and some sharing strategy. - gimel
This seems to fail if, instead of doing 'import config' you do 'from config import x'; the changes to x in one module importing config are not visible to other modules importing config. Does anyone understand why? - BlueBomber
(2) @BlueBomber: because x would be a local identifier to the module that imported it. Think this way: a = SomeClass(); x = a.someattrib; x = 10. Would you expect a's values to have changed? No, right? Same with modules. - MestreLion
(2) @BlueBomber: think of modules as objects (which they are). Importing is just assigning them to a local identifier. from config import x means, in layman's terms, the same as x = config.x. But be aware that if x is mutable (like a list), appending items to it will reflect in config! - MestreLion
What if is large, going to multiple MPI cores and I only want to import it once and share it with all classes? - jtlz2
@user1021819 Sounds like something you could ask in a separate question.. If your multiple MPI cores don't need to change things in config (and expect other MPI cores to see the changes), you're fine. Otherwise it'll have to be done over MPI somehow, and that rather depends on your setup and will likely have to be coded by you. - drevicko
Intuitively I like this - though I am unsure whether I should. This has the feel of a Singleton class to me. Is that a reasonable way to think of it? - Praxiteles
[+57] [2011-07-12 12:35:08] SingleNegationElimination

Python uses a simple heuristic to decide which scope it should load a variable from, between local and global. If a variable name appears on the left hand side of an assignment, but is not declared global, it is assumed to be local. If it does not appear on the left hand side of an assignment, it is assumed to be global.

>>> import dis
>>> def foo():
...     global bar
...     baz = 5
...     print bar
...     print baz
...     print quux
>>> dis.disassemble(foo.func_code)
  3           0 LOAD_CONST               1 (5)
              3 STORE_FAST               0 (baz)

  4           6 LOAD_GLOBAL              0 (bar)
              9 PRINT_ITEM          
             10 PRINT_NEWLINE       

  5          11 LOAD_FAST                0 (baz)
             14 PRINT_ITEM          
             15 PRINT_NEWLINE       

  6          16 LOAD_GLOBAL              1 (quux)
             19 PRINT_ITEM          
             20 PRINT_NEWLINE       
             21 LOAD_CONST               0 (None)
             24 RETURN_VALUE        

See how baz, which appears on the left side of an assignment in foo(), is the only LOAD_FAST variable.

(4) The heuristic looks for binding operations. Assignment is one such operation, importing another. But the target of a for loop and the name after as in with and except statements also are bound to. - Martijn Pieters
Also see the official tutorial: the execution of a function introduces a new symbol table used for the local variables of the function; all variable assignments in a function store the value in the local symbol table; variable refs are resolved by checking enclosing function symbol tables, then globals and built-ins. - Yibo Yang
[+34] [2009-01-08 09:03:33] J S

If you want to refer to a global variable in a function, you can use the global keyword to declare which variables are global. You don't have to use it in all cases (as someone here incorrectly claims) - if the name referenced in an expression cannot be found in local scope or scopes in the functions in which this function is defined, it is looked up among global variables.

However, if you assign to a new variable not declared as global in the function, it is implicitly declared as local, and it can overshadow any existing global variable with the same name.

Also, global variables are useful, contrary to some OOP zealots who claim otherwise - especially for smaller scripts, where OOP is overkill.

[+24] [2014-07-04 10:23:56] Rauni

In addition to already existing answers and to make this more confusing:

In Python, variables that are only referenced inside a function are implicitly global. If a variable is assigned a new value anywhere within the function’s body, it’s assumed to be a local. If a variable is ever assigned a new value inside the function, the variable is implicitly local, and you need to explicitly declare it as ‘global’.

Though a bit surprising at first, a moment’s consideration explains this. On one hand, requiring global for assigned variables provides a bar against unintended side-effects. On the other hand, if global was required for all global references, you’d be using global all the time. You’d have to declare as global every reference to a built-in function or to a component of an imported module. This clutter would defeat the usefulness of the global declaration for identifying side-effects.

Source: What are the rules for local and global variables in Python? [1].


[+20] [2013-10-03 05:41:16] Bohdan

With parallel execution, global variables can cause unexpected results if you don't understand what is happening. Here is an example of using a global variable within multiprocessing. We can clearly see that each process works with its own copy of the variable:

import multiprocessing
import os
import random
import sys
import time

def worker(new_value):
    old_value = get_value()
    set_value(random.randint(1, 99))
    print('pid=[{pid}] '
          'old_value=[{old_value:2}] '
          'new_value=[{new_value:2}] '

def get_value():
    global global_variable
    return global_variable

def set_value(new_value):
    global global_variable
    global_variable = new_value

global_variable = -1

print('before set_value(), get_value() = [%s]' % get_value())
print('after  set_value(), get_value() = [%s]' % get_value())

processPool = multiprocessing.Pool(processes=5), iterable=range(15))


before set_value(), get_value() = [-1]
after  set_value(), get_value() = [-2]
pid=[53970] old_value=[-2] new_value=[ 0] get_value=[23]
pid=[53971] old_value=[-2] new_value=[ 1] get_value=[42]
pid=[53970] old_value=[23] new_value=[ 4] get_value=[50]
pid=[53970] old_value=[50] new_value=[ 6] get_value=[14]
pid=[53971] old_value=[42] new_value=[ 5] get_value=[31]
pid=[53972] old_value=[-2] new_value=[ 2] get_value=[44]
pid=[53973] old_value=[-2] new_value=[ 3] get_value=[94]
pid=[53970] old_value=[14] new_value=[ 7] get_value=[21]
pid=[53971] old_value=[31] new_value=[ 8] get_value=[34]
pid=[53972] old_value=[44] new_value=[ 9] get_value=[59]
pid=[53973] old_value=[94] new_value=[10] get_value=[87]
pid=[53970] old_value=[21] new_value=[11] get_value=[21]
pid=[53971] old_value=[34] new_value=[12] get_value=[82]
pid=[53972] old_value=[59] new_value=[13] get_value=[ 4]
pid=[53973] old_value=[87] new_value=[14] get_value=[70]

[+17] [2016-01-01 19:55:14] Aaron Hall

If I create a global variable in one function, how can I use that variable in another function?

We can create a global with the following function:

def create_global_variable():
    global global_variable # must declare it to be a global first
    # modifications are thus reflected on the module's global scope
    global_variable = 'Foo' 

Writing a function does not actually run its code. So we call the create_global_variable function:

>>> create_global_variable()

Using globals without modification

You can just use it, so long as you don't expect to change which object it points to:

For example,

def use_global_variable():
    return global_variable + '!!!'

and now we can use the global variable:

>>> use_global_variable()

Modification of the global variable from inside a function

To point the global variable at a different object, you are required to use the global keyword again:

def change_global_variable():
    global global_variable
    global_variable = 'Bar'

Note that after writing this function, the code actually changing it has still not run:

>>> use_global_variable()

So after calling the function:

>>> change_global_variable()

we can see that the global variable has been changed. The global_variable name now points to 'Bar':

>>> use_global_variable()

Note that "global" in Python is not truly global - it's only global to the module level. So it is only available to functions written in the modules in which it is global. Functions remember the module in which they are written, so when they are exported into other modules, they still look in the module in which they were created to find global variables.

Local variables with the same name

If you create a local variable with the same name, it will overshadow a global variable:

def use_local_with_same_name_as_global():
    # bad name for a local variable, though.
    global_variable = 'Baz' 
    return global_variable + '!!!'

>>> use_local_with_same_name_as_global()

But using that misnamed local variable does not change the global variable:

>>> use_global_variable()

Note that you should avoid using the local variables with the same names as globals unless you know precisely what you are doing and have a very good reason to do so. I have not yet encountered such a reason.

[+12] [2009-01-09 11:56:19] Kylotan

You're not actually storing the global in a local variable, just creating a local reference to the same object that your original global reference refers to. Remember that pretty much everything in Python is a name referring to an object, and nothing gets copied in usual operation.

If you didn't have to explicitly specify when an identifier was to refer to a predefined global, then you'd presumably have to explicitly specify when an identifier is a new local variable instead (for example, with something like the 'var' command seen in JavaScript). Since local variables are more common than global variables in any serious and non-trivial system, Python's system makes more sense in most cases.

You could have a language which attempted to guess, using a global variable if it existed or creating a local variable if it didn't. However, that would be very error-prone. For example, importing another module could inadvertently introduce a global variable by that name, changing the behaviour of your program.

[+9] [2013-10-13 16:07:41] user2876408

As it turns out the answer is always simple.

Here is a small sample module. It is is a way to show it in a main definition:

def five(enterAnumber,sumation):
    global helper
    helper  = enterAnumber + sumation

def isTheNumber():
    return helper

Here is a way to show it in a main definition:

import TestPy

def main():
    atest  = TestPy

if __name__ == '__main__':

This simple code works just like that, and it will execute. I hope it helps.

thanks, i'm new to python, but know a bit of java. what you said worked for me. and writing global a<ENTER> within the class.. seems to make more sense to me than within a function writing 'global a'.. I notice you can't say global a=4 - barlop
(1) This is probably the simplest yet very useful python trick for me. I name this module global_vars, and initialize the data in init_global_vars, that being called in the startup script. Then, I simply create accessor method for each defined global var. I hope I can upvote this multiple times! Thanks Peter! - swdev
What if there are many many global variables and I don't want to have to list them one-by-one after a global statement? - jtlz2
[+9] [2014-12-04 06:27:43] gxyd

What you are saying is to use the method like this:

globvar = 5

def f():
    var = globvar

f()**  # Prints 5

But the better way is to use the global variable like this:

globavar = 5
def f():
    global globvar
f()   #prints 5

Both give the same output.

The first code gives a syntax error. - Fermi paradox
[+9] [2014-12-20 12:45:26] Mohamed El-Saka

You need to reference the global variable in every function you want to use.

As follows:

var = "test"

def printGlobalText():
    global var #wWe are telling to explicitly use the global version
    var = "global from printGlobalText fun."
    print "var from printGlobalText: " + var

def printLocalText():
    #We are NOT telling to explicitly use the global version, so we are creating a local variable
    var = "local version from printLocalText fun"
    print "var from printLocalText: " + var

Output Result:
var from printGlobalText: global from printGlobalText fun.
var from printLocalText: local version from printLocalText
[Finished in 0.1s]

(2) 'in every function you want to use' is subtly incorrect, should be closer to: 'in every function where you want to update' - spazm
[+6] [2015-02-04 19:19:54] Sagar Mehta

Try this:

def x1():
    global x
    x = 6

def x2():
    global x
    x = x+1
    print x

x = 5

[+5] [2015-10-24 15:46:18] M Newton

Following on and as an add on, use a file to contain all global variables all declared locally and then 'import as':


Stocksin = 300
Prices = []


import  initval as  iv

Def   getmystocks (): 
     iv.Stocksin  = getstockcount ()

Def getmycharts ():
    For ic in range (0,iv.Stocksin):


What is the advantage to move the global variables to another file? Is it just to group together the global variables in a tiny file? And why using the statement import ... as ...? Why not just import ...? - olibre
Ah... I have finally understood the advantage: No need to use the keyword global :-) => +1 :-) Please edit your answer to clarify these interrogations that other people may also have. Cheers - olibre
[+4] [2016-01-07 20:41:19] Mike Lampton

Writing to explicit elements of a global array does not apparently need the global declaration, though writing to it "wholesale" does have that requirement:

import numpy as np

hostValue = 3.14159
hostArray = np.array([2., 3.])
hostMatrix = np.array([[1.0, 0.0],[ 0.0, 1.0]])

def func1():
    global hostValue    # mandatory, else local.
    hostValue = 2.0

def func2():
    global hostValue    # mandatory, else UnboundLocalError.
    hostValue += 1.0

def func3():
    global hostArray    # mandatory, else local.
    hostArray = np.array([14., 15.])

def func4():            # no need for globals
    hostArray[0] = 123.4

def func5():            # no need for globals
    hostArray[1] += 1.0

def func6():            # no need for globals
    hostMatrix[1][1] = 12.

def func7():            # no need for globals
    hostMatrix[0][0] += 0.33

print "After func1(), hostValue = ", hostValue
print "After func2(), hostValue = ", hostValue
print "After func3(), hostArray = ", hostArray
print "After func4(), hostArray = ", hostArray
print "After func5(), hostArray = ", hostArray
print "After func6(), hostMatrix = \n", hostMatrix
print "After func7(), hostMatrix = \n", hostMatrix

[+1] [2017-04-07 18:52:13] Martin Thoma

In case you have a local variable with the same name, you might want to use the globals() function [1].

globals()['your_global_var'] = 42