persisting data within a module

[Peter J. Bismuti]

I'm having trouble understanding how namespaces work in modules. I want to
execute a module within the interpreter and then have values that are
calculated persist so that other modules that get executed can retrieve them.

For example, consider the two simple modules below. The first method fails
and I'm not sure exactly why. (Note: assume one instance of an interpreter.
In my case a 3rd party software tool that starts an interpreter when it
launches).


Two alternate ways of running it:

1. (FAILS: RESULTS A = 0) Use the module "test" itself as the driver using
the conditional statement if (__name__=="__main__"):

test.py
run2.py

or,

2. (SUCCES: RESULTS A = 10) Use "run.py" as the driver.

run.py




_________test.py__________________


import sys,os

A = 0

def getA():
global A
return A

def run():
global A
A = 10

if (__name__=="__main__"):
run()


_________run.py__________________

import test

test.run()
print "A = " + str(test.getA())



_________run2.py__________________

import test

print "A = " + str(test.getA())

--


Peter Bismuti
Boeing Information Technology
Renton, WA
(425) 234-0873 W
(425) 442-7775 C

 

[davisn90210]

I'm having trouble understanding how namespaces work in modules. I want to
execute a module within the interpreter and then have values that are
calculated persist so that other modules that get executed can retrieve them.

Modules retain state their state across all imports in the same
interpreter instance. Module state is not shared among different
instances of the interpreter.

For example, consider the two simple modules below. The first method fails
and I'm not sure exactly why. (Note: assume one instance of an interpreter.
In my case a 3rd party software tool that starts an interpreter when it
launches).

Two alternate ways of running it:

1. (FAILS: RESULTS A = 0) Use the module "test" itself as the driver using
the conditional statement if (__name__=="__main__"):

test.py
run2.py

Ok, what do you mean by this? Do you mean run test.py and then run
run2.py? In so, then you will have *two* instances -- one for each
file being executed. You can only have one main module per
interpreter instance. I suspect this is the source of your confusion.

or,

2. (SUCCES: RESULTS A = 10) Use "run.py" as the driver.

run.py

_________test.py__________________

import sys,os

A = 0

def getA():
global A
return A

def run():
global A
A = 10

if (__name__=="__main__"):
run()

Here, A is only initialized when the module is loaded iff it is the
main module. If it's not the main module, then it will have A set to
0 until some other code calls run().

_________run.py__________________

import test

test.run()
print "A = " + str(test.getA())

This code calls test.run(), which is necessary for A to be 10.

_________run2.py__________________

import test

print "A = " + str(test.getA())

This code gets the value of test.A without calling test.run(). Since
test.run() was not called, A is the value it was initialized when the
test module was loaded -- namely, 0.

Hope this helps,

--Nathan Davis

 

[Peter J. Bismuti]

I'm not sure how to better state my question than to post my code.

The question boils down to which namespace to variable in the module (in this
case A) end up in depending on whether or not the module is simply imported
by another module which acts as the driver (run.py is __main__), or when the
module drives itself (test.py __main__).

For example: my guess is that in the first case A ends up in the namespace of
test and could be referenced by test.A, in the second case A ends up in the
global namespace (and therefore must be referred to as simply A by other
modules).

Can anyone please shed some light on this for me?

Thanks

Modules retain state their state across all imports in the same
interpreter instance. Module state is not shared among different
instances of the interpreter.


Ok, what do you mean by this? Do you mean run test.py and then run
run2.py? In so, then you will have *two* instances -- one for each
file being executed. You can only have one main module per
interpreter instance. I suspect this is the source of your confusion.


Here, A is only initialized when the module is loaded iff it is the
main module. If it's not the main module, then it will have A set to
0 until some other code calls run().


This code calls test.run(), which is necessary for A to be 10.


This code gets the value of test.A without calling test.run(). Since
test.run() was not called, A is the value it was initialized when the
test module was loaded -- namely, 0.

Hope this helps,

--Nathan Davis

--


Peter Bismuti
Boeing Information Technology
Renton, WA
(425) 234-0873 W
(425) 442-7775 C

 

[Peter J. Bismuti]

How is that state different depending on whether a module has been simply
imported (#2. some other block of code has __name__ == "__main__") and the
script itself being run (#1. and having __name__=="__main__")?

Ultimately, what I want is for a module to remember (persist) the value of A,
regardless of how the module has been loaded into the interpreter.

Thanks

Modules retain state their state across all imports in the same
interpreter instance. Module state is not shared among different
instances of the interpreter.


Ok, what do you mean by this? Do you mean run test.py and then run
run2.py? In so, then you will have *two* instances -- one for each
file being executed. You can only have one main module per
interpreter instance. I suspect this is the source of your confusion.


Here, A is only initialized when the module is loaded iff it is the
main module. If it's not the main module, then it will have A set to
0 until some other code calls run().


This code calls test.run(), which is necessary for A to be 10.


This code gets the value of test.A without calling test.run(). Since
test.run() was not called, A is the value it was initialized when the
test module was loaded -- namely, 0.

Hope this helps,

--Nathan Davis

--


Peter Bismuti
Boeing Information Technology
Renton, WA
(425) 234-0873 W
(425) 442-7775 C

 

[Gabriel Genellina]

En Tue, 13 Nov 2007 13:09:01 -0300, Peter J. Bismuti

How is that state different depending on whether a module has been simply
imported (#2. some other block of code has __name__ == "__main__") and
the
script itself being run (#1. and having __name__=="__main__")?

It's not different at all, or I don't understand the question.

Ultimately, what I want is for a module to remember (persist) the value
of A,
regardless of how the module has been loaded into the interpreter.

It doesn't care. If you have a variable A in (the global namespace of) a
module, it's there, no matter how the module has been loaded. A namespace
has no concept of "history", it's just a mapping from names to objects.

Unless you're talking about this situation (should be on the FAQ, but I
can't find it):

--- begin one.py ---
A = 1

if __name__=='__main__':
print "In one.py, A=", A
import two
print "In one.py, after importing two, A=", A
--- end one.py

--- begin two.py ---
import one
print "In two.py, one.A=", one.A
one.A = 222
print "In two.py, after modifying one.A=", one.A
--- end one.py

Executing:

>python one.py

you get this output:

In one.py, A= 1
In two.py, one.A= 1
In two.py, after modifying one.A= 222
In one.py, after importing two, A= 1

In this (pathological) case, there are TWO different instances of the
one.py module, because modules are indexed by name in the sys.modules
dictionary, and the first instance is under the "__main__" name, and the
second instance is under the "one" name.
So: don't import the main script again.