python first assignment of a global variable

[Rodrigue]

Hi all,

I came accross a strange behaviour in python today. Here is a simple
example to describe my situation:

MY_GLOBAL = ''

def a():
print 'global is: ', MY_GLOBAL

def b():
try:
MY_GLOBAL += 'bla'
except Exception, e:
print 'b: ', e

def c():
try:
global MY_GLOBAL
MY_GLOBAL += 'bla'
except Exception, e:
print 'c: ', e

def d():
try:
if not MY_GLOBAL:
print 'not my global'
except Exception, e:
print 'd: ', e

def e():
try:
if not MY_GLOBAL:
MY_GLOBAL = ''
except Exception, e:
print 'e: ', e

def f():
global MY_GLOBAL
if not MY_GLOBAL:
MY_GLOBAL = ''

def e_raise():
if not MY_GLOBAL:
MY_GLOBAL = 'bla'

if __name__ == "__main__":
a()
b()
c()
d()
e()
f()
e_raise()


And here is the output I get:

global is:
b:  local variable 'MY_GLOBAL' referenced before assignment
e:  local variable 'MY_GLOBAL' referenced before assignment
Traceback (most recent call last):
  File "glo.py", line 49, in 
e_raise()
  File "glo.py", line 39, in e_raise
if not MY_GLOBAL:
UnboundLocalError: local variable 'MY_GLOBAL' referenced before
assignment


Now, I was reading that page 
http://stackoverflow.com/questions/370357/python-variable-scope-question
and found (understood) only part of the behaviour that could explain
the output.

Basically, I was very surprised to discover that e() raises an
exception, but even more that e_raise() points to
if not MY_GLOBAL Is the problem not really when I assign?

My assumption is that some reordering is happening behind the scenes
that creates a situation similar to the += which assigns hence expects
to be at the local level.

I would love some enlightenment here

Rodrigue
-- 
http://mail.python.org/mailman/listinfo/python-list

Re: python first assignment of a global variable

[Rodrigue]

> MY_GLOBAL, by virtue of being assigned to later in the function, and the
> absence of a global statement, is identified as a local variable.

> When a function contains a
> single assignment (or augmented assignment) to a name, the compiler
> generates bytecode such that all references to that name within the
> function will be looked up in the local scope only

Alright. I didn't know that. I thought the entire scope (local, then
global) was considered in every situation. It does explain the
observed behaviour then.

I'm surprised I never bumped into that before, but I'm glad I learnt
something new about python today.

Thanks Emile and Miles for the explanation!
-- 
http://mail.python.org/mailman/listinfo/python-list