Isn't this a bug?
Here's the test program:
import code
def test_func():
lv = 1
print '\n\nBEFORE lv: %s\n' % (lv)
code.interact(local=locals())
print '\n\nAFTER lv: %s\n' % (lv)
return
test_func()
gv = 1
print '\n\nBEFORE gv: %s\n' % (gv)
code.interact(local=locals())
print '\n\nAFTER gv: %s\n' % (gv)
Here's the output and interactive session:
BEFORE lv: 1
Python 2.4 (#60, Nov 30 2004, 11:49:19) [MSC v.1310 32 bit (Intel)] on win32
Type "help", "copyright", "credits" or "license" for more information.
(InteractiveConsole)
>>> lv = 2
>>> print lv
2
>>> ^Z
AFTER lv: 1
BEFORE gv: 1
Python 2.4 (#60, Nov 30 2004, 11:49:19) [MSC v.1310 32 bit (Intel)] on win32
Type "help", "copyright", "credits" or "license" for more information.
(InteractiveConsole)
>>> gv = 2
>>> print gv
2
>>> ^Z
AFTER gv: 2
In the case of code.interact being called inside the function, the locals
were available to the interactive environment but changes did not stick when
you returned back the function. (started as lv=1 changed to lv=2 in
interactive session, back to lv=1 in function) Since you are still in the
function and since code.interact used the same local environment why didn't
the change stick until testfunc ended?
When code.interact was called from the outside the function (globals() ==
locals()) the changes stuck. (started as gv=1, changed to gv=2 in
interactive session, stuck as gv=2 back in main).
"Steve Holden" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]
> Michael Hoffman wrote:
>> Joe wrote:
>>
>>> I want the script to decide whether to fall back to the interactive
>>> prompt. You solution makes it ALWAYS fall back to the interactive
>>> prompt.
>>
>>
>> Actually, using sys.exit() means the program can exit even if python -i
>> is used.
>>
>> You can use:
>>
>> import code
>> code.interact()
>>
>> which emulates the interactive prompt.
>
> Unfortunately it does so in an entirely new namespace, thereby losing the
> advantage of -i - namely, that you can investigate the program's namespace
> after it's terminated.
>
> regards
> Steve
>
--
http://mail.python.org/mailman/listinfo/python-list
